This page is located as http://www.ida.liu.se/~pelab/ALModelica
Project has started the 2nd of April 2001.
The use of computer
simulation in industry is rapidly increasing. Simulation is typically used to
optimize product properties and to reduce product development cost and time to
market. Whereas in the past it was considered sufficient to simulate subsystems
separately, the current trend is to simulate increasingly complex physical
systems composed of subsystems from multiple application domains such as
mechanical, electric, hydraulic, thermodynamic, control system components and
material property descriptions. Even models in a single application domain tend
to become more complex and they require high flexibility in modeling.
In order to
efficiently create and maintain complex simulation software, the problems should
be stated at a high level of abstraction, i.e. as object-oriented mathematical
model. For this purpose a high level of model description should use the latest
developments in computer language design. High performance of numerical
simulation tools is necessary. To achieve high performance it is necessary to
use the best translation, parallellization, and code generation techniques. In
order to analyze simulation results, a powerful and flexible interactive problem
solving environment is necessary, including animation and visualization tools.
For this purpose the
research programme of this application integrates two major parts:
Methods and
tools for study of computational materials science.
General methods
and tools for study of complex physical systems, with applications to
computational material science.
Constructing
a general-purpose, high-level integrated modeling, simulation and
visualization environment. This environment can serve as a toolbox for simulation applications in many different areas. In
particular, such tool components (in form of object-oriented libraries) will
be constructed, that support computations for multi-domain physics including
the computational materials science. This environment will be applicable in
industry R&D as well as university research and education.
Modeling and
simulation will have an increased role in the development and use of industrial
processes and products. However, widely
available industrial use of simulation requires easily configured and validated
models, the establishment of tailored model libraries, combination of both
multidisciplinary and multi-vendor tools, and much faster model development and
easier experimentation than presently possible. Fast prototyping by simulation
shortens commissioning and production times and increases quality.
Industrial partners
from several application domains have expressed a strong interest in fast
prototyping of high quality models spanning the whole lifecycle from early
design phases to training simulators. Current simulation tools don't allow
multi-layered structuring that has proven enormously effective in the software
industry. Modelica can combine engineering
modeling expertise with the high increase in productivity of modern CASE tools
to form a new standard in the currently fragmented simulation software industry.
This way, industrial engineers can use packaged knowledge incorporated in the
models from modeling specialists coming from research centers or universities.
The second applicant (Dr. Vadim Engelson) is a member of the Programming
Environments Group (PELAB) at IDA, involved in this area. This group has strong cooperation
with SKF (Bearing simulation group led by Prof. Dag Fritzson; the
cooperation includes modeling, high performance parallel simulation and
visualization of mechanical models), MathCore
AB (development of the Modelica language and tools), Dynasim
AB (development of Modelica tools and libraries), CyCore
AB (development of visualization tools). PELAB has also strong international
cooperation including several European industrial partners (DLR, ABB,
KUKA-Roboter etc.) in the RealSim
project.
Understanding solid materials behavior during synthesis and
operation are potentially crucial steps in the advancement of key Swedish
industries such as electronics and engineering. Better knowledge of these
atomistic processes has allowed the control of deposition parameters in the
manufacture of high-density ICs tailored for applications in, for example,
information technology. The first applicant (Doc. Valeriu Chirita) is a member
of the Thin Film group at IFM, currently involved in fundamental (experimental)
research in these areas (http://www.ifm.liu.se/thinfilm).
The group has strong connections with numerous industrial organizations such as ABB,
SECO Tools, SKF, which have strategic interests
in developing new, high performance materials. To achieve this goal, a
general-purpose high-level computational software package, dedicated to
materials related problems, would be instrumental in correlating experimental
and theoretical research, within university and industrial environments.
Furthermore, the
high level of re-usability allows for building models better suited for the
modeling goal. Another issue important for industrial end users is the
availability of a fully spread set of component models that can be easily
connected together to obtain more complex system models.
Today, scientific
programmers face the difficult task of choosing between middle-level languages,
well known for their object-oriented capabilities, and low-level languages,
which offer better efficiency in numerical calculations. The former class
includes languages such as Java (with obvious Internet applicability) and C++,
while Fortran is the typical example for the latter.
The challenge
is to make an efficient transition from numerical computation to
object-orientation in multiple domains, using high-level languages, like
Modelica, in materials related problems. With appearance of multi-domain
high-level modelling and simulation methodologies and tools it becomes possible
to develop reusable libraries of components for many new application areas,
which earlier could use just specialized software packages. As we are entering
the era of applied quantum mechanics in materials science, the major task is to
extend multi-domain features of Modelica or similar methodologies and to develop
libraries of components, which ultimately will be used in computational material
science.
Make synergy
between Molecular Dynamics and Modelica.
The research will
concentrate in developing, improving, and testing the computational tools
to be incorporated in the high-level software package. At the same time the
features of Modelica that are immediately applicable to modeling in material
science will have to be identified. The main goal of this period is to
make a successful incorporation of existing software components into the
multi-domain environment of Modelica. Tests will be carried out on a variety of
systems, with emphasis on technologically relevant materials (e.g. ceramics and
metal alloys). This will be done for given phenomena and model materials.
The milestone
of this project is to construct a high-level software tool, which in the final
form, will have the following features:
State-of-the
art - interaction potentials are at the heart of any computer simulation
related to materials science, and as such, novel, improved functionals need
be included to allow large-scale simulations, of a wide range of materials.
Operational
versatility – allow the expert user to design and test new interaction
potentials, to interface with existing commercial software packages. This
would ensure that the package is also future-proof.
User-friendly
interface & portability – currently most ab-initio and classical,
MC and/or MD programs are in-house developments, and as such, their
portability from platform to platform is cumbersome. This feature would be
extremely useful for expert and non-expert users. A plug-and-play modeling
paradigm supported in Modelica graphical modeling tools will be used.
Visualization
capability – if integrated, these tools are essential in analyzing
results, play back of simulations in movie format and development of
Internet compatible applications. This would also increase the potential for
use in teaching and distributed (Internet-based) engineering.
The Modelica-based
environment should become an integrated object-oriented problem-solving
environment also for computational material science.
Continue
incorporating breakthroughs in computational methods in the package.
Develop
application specific modules to address the needs of different industries.
Integrate the
package for general teaching purposes at LiU.
Seek industrial
partners to ensure the financial stability of the project.
Search avenues
for possible marketing in R&D and/or teaching.
The CENIIT support
is necessary for Vadim Engelson in order to establish his position as a research
leader, provide adequate research leadership in the Object-oriented simulation
and visualization subgroup, and supervise the PhD students in their research.
Currently, the group
of people working on materials related computer simulations at LiU is quite
small, basically just Doc. V. Chirita and Doc. P. Münger. In the last few
years, the MD simulations group in Linköping has become one of the most active,
on the international scale, in the theoretical study of atomic scale processes
during early stages of epitaxial growth. The group has now reached a critical
mass and support for this project is necessary for establishing a research group
to work on integrating developments in information technology into research in
materials related problems. As part of this project, a PhD student, or
postdoctoral associate would be included in the group, working primarily on
developing the object-oriented high-level software package.
The CENIIT support would offer the possibility for long term planning and the opportunity to establish research in a new field at LiU, which is complementary to present interests, has significant industrial relevance, and is of mutual benefit to the groups involved. The project will also be instrumental in fostering the co-operation between several groups at IFM (modelling, theory and experimental groups) and the Programming Environments Laboratory (PELAB) at IDA.
Updated
on Friday, April 06, 2001
by Vadim Engelson