826
Proceedings of the 18
th
International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
2.2
Unsolved problems of soil mechanics
Imperfection of models representing soil behaviour. It seems
that there are dozens of models of soil-behaviour, however,
often, in real life one is faced with situations of being unable to
find a requisite model capable of effective and correct solution
of a practical problem.
Insufficiency of initial input parameters submitted as results
of site investigation.
Limitations of models and dearth of good quality input
parameters often yields imprecise modelling results.
2.3
Organizational problems
Absence of unified data base on in situ monitoring of buildings
and structures. Improvement of models and practice of analyses
is impossible without comparison with in situ monitoring data.
Nevertheless, despite many construction projects built all
around the world, there are only single instances of well
documented in situ monitoring.
Sharing responsibility between geotechnical and super-
structure engineers.
3 NEW RELEASE OF FEMMODELS SOFTWARE
(FEM MODELS 3.0) AS A POSSIBLE INSTRUMENT OF
SSI PROBLEMS SOLUTIONS
The list of problems necessary to be solved in order to enhance
SSI reality is a heavy one. Is there a possibility to overcome
them?
Such instrument can be found in the new release of
FEM Models 3.0
software. Technical SSI problems can
obviously be resolved by creating an efficient programme code
capable of rapid SSI problems solutions. Additionally, creating
an efficient software medium for such analyses can remove a
number of other problems. For example, in when a convenient
programme capable of catering to the interests of both
superstructure and geotechnical engineers equally well is
created organizational problems listed above are largely
removed, because division of labour and sharing responsibility
can be realized within the framework of a single analysis profile
“soil-structure”.
The new release is being created under the auspices of
ISSMGE TC 207 “Soil-Structure Interactions and Retaining
Walls”. The larger portion of the software is structured as an
“open source”, involving maximum openness and availability.
3.1.
3-D modelling medium
. The medium is created based on
a very effective open software source OpenCascade, capable of
performing solid modelling using logical operations with
figures, their geometric transformations, etc. The profile editor
contains rather simple modelling instruments, perfectly
accessible to engineers who are the end-users of the software.
An important task is introduction of adaptive finite-element
meshes into the modelling medium architecture. Adaptive
meshes move a significant portion of the task to provide
solution accuracy to the computation algorithm.
3.2.
Non-linear equations solver
. When using finite element
method, non-linear differential equations within the limits of
investigated areas are reduced to solution of systems of non-
linear algebraic equations. The task of this software component
is an effective solution of large systems of nonlinear equations.
3.3.
Integrated Engineering Environment (IEE)
provides a
possibility to perform engineering computation (in the form of
formulas and simple algorithms), as well as solution of
parametric problems by finite elements methods using
components 2.1 and 2.2 described above. IEE is a
java
based
tool and provides all possibilities of a state-of-the-art high-level
programming language aimed at simplifying writing of
mathematical formulae and algorithms (work with matrices,
numerical modelling and so on). The codes of finite-element
models and materials models are written in the same
environment in the most transparent and accessible form. The
objective of this environment is to make complex non-linear
models more accessible for study and improvement.
3.4.
Library of finite elements and materials models
. All
models of elements and materials are stored in the library with
an open source, which makes their analysis and verification
easier. This library makes it possible not only to use preset finite
elements solutions, but also to add some of one’s own design.
3.5.
Library of parametric problems to be used in design
practice
. This
library provides a possibility for engineers to
solve specific practical problems without an in-depth study of
finite-element programmes. For instance, the way SSI analyses
are performed by the authors is such that numerical analyses are
always correlated with available analytical solutions, which
yield approximated results. Such approach eliminates potential
for significant errors. Using the library of parametric problems
the user is given a possibility, for instance, to make both an
ordinary analytical calculation of settlements and a numerical
computation, following which both results can be collated.
All parts of the programmes (except the solver) are of open
source type. To ensure a fastest possible solution it is suggested
to use a special highly effective server. To increase computation
quality, collection of data on well-documented case histories is
actively underway with participation of ISSMGE TС 207 «Soil
Structure Interaction». In future based on this work it is planned
to build a method of testing soil-models as regards their
correspondence to in situ data.
4 USE OF NON-LINEAR SOIL MODELS FOR
MODELLING SUBSOIL IN SSI APPLICATIONS
The issues of adequate choice of soil model to properly
represent subsoil action are covered in every detail in the paper
(Shashkin, 2010a).
The most promising approach to building the model in our
opinion is the so called Double Hardening, which perhaps
would better be referred to as
Independent Hardening
. In this
approach hardening zones during isotropic and deviatoric
loading develop independently, which is confirmed
experimentally. An example of the model, built based on that
principle is he Hardening Soil Model (HSM). However, the idea
of independent hardening in it is not brought to its logical
conclusion.
Additionally, the standard HSM model assumes isotropic
hardening, which can not be proved experimentally. At
multidirectional, specifically, cyclic loading (Fig. 2) the model
yields results radically different from the experiments.
Fig. 2. Strain-load dependency at cyclic loading: 1 – as evidenced by
experiments; 2 – according to isotropic hardening models
To remove the above-described discrepancies the present
authors developed their own model featuring independent
anisotropic hardening. Compared with the HSM model the
dependencies for hardening in volumetric and deviatoric
-0.2
-0.1
0
0.1
0.2
40
20
20
40
1
2
