The 48th Scandinavian Conference on
Simulation and Modeling (SIMS 2007)
30-31 October 2007, Gothenburg, Särö.
Final Program
Program Co-Chairs
Peter Bunus
Department of Computer and Information Science
Linköping University, Linköping, Sweden
Dag Fritzson
SKF, Göteborg and Department of Computer and Information Science
Linköping University, Linköping, Sweden
Claus Führer
Center for Mathematical Sciences
Lund University, Lund, Sweden
Program Committee
Felix Breitenecker Felix.Breitenecker@tuwien.ac.at
Vienna Univ. of Technology, Inst. f. Analysis and Scientific Computation RG Mathematical
Modelling and Simulation
Austria
Erik Dahlquist, erik.dahlquist@mdh.se
Mälardalens Högskola, Sweden
Brian Elmegaard, be@mek.dtu.dk
Department of Energy Engineering, Technical University of Denmark
Hilding Elmqvist
Dynasim AB , Lund, Sweden,
Peter Fritzson
Department of Computer and Information Science
Linköping University, Sweden
Esko Juuso, esko.juuso@oulu.fi
University of Oulu, Finland
Kaj Juslin, kaj.juslin@vtt.fi
VTT Technical Research Centre of Finland, Finland
Bernt Lie, Bernt.Lie@hit.no
Høgskolen i Telemark, Norway
Per Lidström
Lund University, Division of Mechanics, Department of Mechanical Engineering, Sweden
2
Time
08:00
09:50
09:50
10:00
10:00
Tuesday 30 October
(Peter Fritzson), Peter Bunus, Adrian Pop
(Department of Computer and Information Science
Linköping University, Sweden )
TUTORIAL: Introduction to Object-Oriented Modeling and Simulation with Modelica
BREAK
Andreas Heckmann
(Deutsches Zentrum für Luft- und Raumfahrt e.V. (DLR), Institute of Robotics and Mechatronics)
TUTORIAL: Modelling and Simulation of Rigid and Flexible Multibody Systems in Modelica
12:00
12:00
13:00
13:00
13:10
13:10
14:00
LUNCH
Opening. Welcome SIMS 2007
KEYNOTE: Gustaf Soderlind
Numerical Analysis, Centre for Mathematical Sciences, Lund University, Sweden
Recent Advances in Adaptive Time-Stepping
SESSION 1. Session Chair: Claus Führer
14:10 Håkan Lundvall and Peter Fritzson. Automatic Parallelization of Mathematical Models Solved with
14:35 Inlined Runge-Kutta Solvers
14:35
Erik Dahlquist. How to communicate in a more efficient way over long distances instead of travelling
15:00
15:00
BREAK + EXHIBITION
15:10
SESSION 2. Session Chair: Esko Juuso
15:10 Kristoffer Norling, David Broman, Peter Fritzson, Alexander Siemers and Dag Fritzson. Secure
15:35
15:35
16:00
16:00
16:25
16:25
16:50
16:50
17:15
17:15
17:40
17:40
18:30
19:00
22:00
Distributed Co-Simulation over Wide Area Networks.
Dag Fritzson, Jonas Ståhl and Iakov Nakhimovski. Transmission line co-simulation of rolling bearing
applications
Claude Lacoursiere. Regularized, stabilized, variational methods for multibodies
Klas Modin, Dag Fritzson and Claus Führer. Adaptive numerical integration of complex multibody problems with application to rolling bearing simulations
Antonio Vecchio. A Mixed Test Multi Body Modelling Approach to Load Simulation on Aircraft Components
Rostyslav Stolyarchuk. Multibody Modeling and Dynamical Analysis of the Worm Gear Drive System
EXHIBITION
CONFERENCE DINER
(Invited Speaker Sten Henriksson - Fifty years with computers.
A talk with a mixture of personal recollections and reflexions on the intellectual contributions of computers to science. Sten Henriksson is a semiretired senior lecturer at the Computer Science department of Lund University. Starting as a programmer with the SMIL computer, he got his Ph.D. in numerical analysis (approximation theory). His interest later turned to human-computer interaction.
3
Time
08:15
08:45
09:00
09:50
09:50
10:00
Wednesday 31 October
ANNUAL SIMS MEETING
KEYNOTE: Pieter J. Mosterman
Senior Research Scientist, The MathWorks, Inc Natick, MA, USA
Model-Based Design at an Enterprise Level: New directions in Modeling and Simulation Research
BREAK + EXHIBITION
SESSION 3.1 Session Chair: Bernt Lie
10:00 Kourosh Mousavi Takami, Örjan Danielsson and
Jafar Mahmoudi. Simulation and optimization of
high power super heater reflectors usable in electrical furnaces for heat loss reduction
10:25
10:25 Lars Øi. Aspen HYSYS Simulation of CO2 Removal by Amine Absorption from a Gas Based
10:50 Power Plant
10:50 Dietmar Winkler and Clemens Gühmann. Modelling of Electrical Faults in Induction Machines Us11:15 ing Modelica
11:15 Mats G Larson and Håkan Jakobsson. Adaptive
Reduction of Finite Element Models of Complex
11:40 Mechanical Components
11:40 Michael Brown and Mark Jones (Bipin Patel presenter). Reconstruction of a Rollover Ejection
Case Using Vehicle Dynamics and Occupant
12:05 Simulation with Human Body Models.
12:05
13:00
SESSION 3.2 Session Chair: Jafar Mahmoudi
Anna Nyström. Modeling and simulation of a semibatch reactor
Adrian Pop, Vasile Baluta and Peter Fritzson.
Eclipse Support for Design and Requirements Engineering based on ModelicaML
Esko Juuso. Intelligent modelling of a fluidised bed
Granulator used in production of pharmaceuticals
Alexander Siemers and Dag Fritzson. A MetaModeling Environment for Mechanical System CoSimulations
LUNCH + SIMS BOARD MEETING
SESSION 4.1 Session Chair: Bipin Patel
SESSION 4.2 Session Chair: Erik Dahlquist
Juan Ignacio Videla and Bernt Lie. Parameter estimation of a small-scale CHP model
Kourosh Mousavi Takami, S.M.Esmail Mousai Takami and Jafar Mahmoudi. Simulation of energy in
the building and design a new intelligent building
with controllable and wise devices
13:00
13:25
13:25 Karin Kraft, Stig Larsson and Mathias Lidberg.
Kourosh Mousavi Takami. Identification of a best
Using an adaptive FEM to determine the optimal
thermal formula and model for oil and winding of
control of a vehicle during a collision avoidance
power transformers using predeiction method
13:50 manoeuvre
13:50 Adrian Pop and Peter Fritzson. Towards Run-time Dmitrii Silvestrov, Anatoliy Malyarenko and Evelina
Debugging of Equation-based Object-oriented
Silvestrova. Stochastic simulation of insurance
Languages
business
14:15
14:15
Per-Olof Johansson, Patrick Ljunggren and Janusz
Tuomas Kataja and Yrjö Majanne. Dynamic
Wollerstrand. Modelling Space Heating Systems
Connected to District Heating in Case of Electric
model of a bubbling fluidized bed boiler
Power Failure
14:40
14:40
BREAK, CLOSING SIMS 2007
15:00
4
Tuesday 31st of October
08:00 – 09:50 Tuesday
Tutorial 1: Introduction to Object-Oriented Modeling
and Simulation with Modelica
Peter Fritzson, Peter Bunus, Adrian Pop
Department of Computer and Information Science,
Linköping University, Sweden
Object-Oriented modeling is a fast-growing area of modeling and simulation that provides a structured, computer-supported way of doing mathematical and equation-based modeling. Modelica is
today the most promising modeling and simulation language in that it effectively unifies and generalizes previous object-oriented modeling languages and provides a sound basis for the basic concepts. The Modelica modeling language and technology is being warmly received by the world
community in modeling and simulation with major applications in virtual prototyping. It is bringing
about a revolution in this area, based on its ease of use, visual design of models with combination of
lego-like predefined model building blocks, its ability to define model libraries with reusable components, its support for modeling and simulation of complex applications involving parts from several application domains, and many more useful facilities.
The tutorial presents an object-oriented component-based approach to computer supported
mathematical modeling and simulation through the powerful Modelica language and its associated
technology. Modelica can be viewed as an almost universal approach to high level computational
modeling and simulation, by being able to represent a range of application areas and providing general notation as well as powerful abstractions and efficient implementations.
The tutorial gives an introduction to the Modelica language to people who are familiar with basic
programming concepts. It gives a basic introduction to the concepts of modeling and simulation, as
well as the basics of object-oriented component-based modeling for the novice, and a an overview of
modeling and simulation in a number of application areas. The OpenModelica environment together
with the graphical user interface MathModelica Lite will be used for hands-on exercises.
The tutorial has several goals:
•
•
•
•
2
Being easily accessible for people who do not previously have a background in modeling and
simulation.
Introducing the concepts of physical modeling, object-oriented modeling and componentbased modeling and simulation.
Demonstrating modeling examples from several application areas.
Providing opportunity for hands-on exercises with the OpenModelica open-source implementation of Modelica and the MathModelica Lite graphic user interface.
10:00 – 12:00 Tuesday
Tutorial 2: Modelling and Simulation of Rigid and Flexible Multibody Systems in Modelica
Andreas Heckmann
Deutsches Zentrum für Luft- und Raumfahrt e.V.
(DLR) Institute of Robotics and Mechatronics
Quite often the mechanical components are the core elements of a complex technical system. Therefore a modelling language such as Modelica relies on the capability to systematically treat the dynamic behaviour of inteconnected bodies influenced by various physical quantities. In order to answer this purpose the Modelica Multibody Library and the Modelica FlexibleBodies Library provide
a range of modeling elements to describe rigid or flexible bodies respectively which may undergo
large 3-dimensional translational and rotational displacements.
The tutorial will give an introduction to these capabilities for people who have a back-ground in
classical engineering mechanics, but not necessarily in multibody dynamics. In particular the goals
of the tutorial are:
•
•
•
•
To present the main modelling components of both libraries from the user’s point of view.
To provide initial hands-on experience.
To describe the main underlying concepts and their theoretical background.
To discuss essential details of the implementation.
As a common platform for exercises, software with both libraries and a test version of the simulation
environment Dymola will be provided (MS Windows operating system). Please bring a laptop with
CD-reader in order to participate in the exercises.
Lecturer: Andreas Heckmann graduated in Mechanical Engineering at the Technical University of
Munich in 1999. He joined the German Aerospace Center, DLR, in 2000 and worked on simulation
method enhancements with the specialised multibody code Simpack that has initially been deveoped
by the DLR’s Vehicle System Dynamics group. In 2005, Andreas Heckmann received his doctor’s
degree at the University of Hannover submitting a thesis on multifield problems such asthermoelasticity in multibody dynamics. Dr. Heckmann is the main author of the Modelica FlexibleBodies Library and contributes multifield capabilities to the ITEA 2 EuroSsLib-Project.
3
13:10 – 14:00 Tuesday
KEYNOTE: Recent Advances in Adaptive Time-Stepping
Gustaf Söderlind
Numerical Analysis, Centre for Mathematical Sciences
Lund University, Sweden
Abstract: Adaptive methods are of cruical importance to efficiency in scientific computing. However, it is still very common that strategies for adaptation used in practice are heuristic or less well
supported by analysis. In this talk we review some advances with a focus on time-stepping, where
we show that ideas from control theory and digital signal processing can be used to construct efficient techniques for adaptive time-stepping, and that these techniques have an impact on the regularity of the computed solutions as well as on the stability of the computed results. If time permits, we
will also have a look at new adaptive methods for Hamiltonian problems, which require that the
adaptivity is time-reversible and symmetric, without losing important conservation properties and
long-time behavior.
4
SESSION 1:
Session Chair: Claus Führer
14 :10 – 14 :35 Tuesday
Automatic Parallelization of Mathematical Models Solved with
Inlined Runge-Kutta Solvers
Håkan Lundvall and Peter Fritzson
PELAB – Programming Environment Lab, Dept. Computer Science
Linköping University, S-581 83 Linköping, Sweden
{haklu, petfr}@ida.liu.se
In this work we report preliminary results of automatically generating parallel code from equationbased models together at two levels: Performing inline expansion of a Runge-Kutta solver combined
with finegrained automatic parallelization of the resulting RHS opens up new possibilities for generating high performance code, which is becoming increasingly relevant when multi-core computers
are becoming common-place.
We have introduced a new way of scheduling the task graph generated from the simulation problem
which utilizes knowledge about locality of the simulation problem. The scheduling is also done in a
way that limits communication, to the greatest extent possible, to neighboring processors thus avoiding expensive global synchronization. Preliminary tests on a PC-cluster show speedup that is better
than what was achieved in previous work where parallelization was done only at the equation system
level.
14 :35 – 15 :00 Tuesday
How to communicate in a more efficient way over long distances instead of travelling
Erik Dahlquist
Malardalen University
erik.dahlquist@mdh.se
Today many long trips are made because of a need to meet physically. This is especially the case for
business travels. If it would be possible to have a long distance contact where it felt like the counterpart was sitting next to you, the need for the actual travel would decrease significantly. This would
save both time and energy. In this paper a discussion about technical possibilities to do this is discussed. A review is made over previous research related to the subject. Also ideas are presented on
how this could be performed using a simulator model that drives a “mechanical, elastic head” from
the information received on-line from a photo taken continuously by the counterpart.
5
SESSION 2:
Session Chair: Esko Juuso
15 :10 – 15 :35 Tuesday
Secure Distributed Co-Simulation over Wide Area Networks
Kristoffer Norling*, David Broman*, Peter Fritzson*, Alexander Siemers**, Dag Fritzson**
*Department of Computer and
Information Science, Linköping University
SE-581 83 Linköping, Sweden
{x06krino,davbr,petfr}@ida.liu.se
**SKF Engineering Research Centre
MDC, RKs-2, SE-415 50 Göteborg, Sweden
alexander.siemers@skf.com
dag.fritzson@skf.com
Modeling and simulation often require different tools for specialized purposes, which increase the
motivation to use co-simulation. Since physical models often are describing enterprises' primary
know-how, there is a need for a sound approach to securely perform modelling and simulation. This
paper discusses different possibilities from a security perspective, with focus on secure distributed
co-simulation over wide area networks (WANs), using transmission line modeling (TLM). An approach is outlined and performance is evaluated both in a simulated WAN environment, and for a
real encrypted co-simulation between Sweden and Australia. It is concluded that several parameters
affect the total simulation time, where especially the network delay (latency) has a significant impact.
15 :35 – 16 :00 Tuesday
Transmission line co-simulation of rolling bearing applications
Dag Fritzson, Jonas Ståhl and Iakov Nakhimovski
SKF Engineering Research Centre, Modelling & Simulation, Göteborg, Sweden
dag.fritzson@skf.com
Connectivity, i.e., connecting different types of simulation tools, becomes increasingly important.
The objective is to be able to do cost effective simulations of a large system by utilizing existing
tools together. The TLM (Transmission Line Modelling) co-simulation technique is stable, does not
have numerical problems, and is proven. It has potential for being a cosimulation technique of general usage. The aim of the paper is an evaluation of TLM technique for rolling bearing applications
via a study of a grinding spindle model. The conclusion is that the results of TLM and non-TLM
simulations are similar for a realistic application with rolling bearings and applicable for engineering
purposes. The numerical stability and performance are satisfactory for the considered application.
6
16 :00 – 16 :25 Tuesday
Regularized, Stabilized, Variational Methods for Multibodies.
Claude Lacoursière
HPC2N/VRlab and Department of Computing Science.
Umeå University. SE-901 87, Umeå, Sweden
claude@hpc2n.umu.se
A time-discrete formulation of the variational principle of mechanics is used to construct a novel
first order, fixed time step integration method for multibody systems subject to mixed constraints.
The new stepper, coined Spook, includes physics motivated constraint regularization and stabilization terms. The stepper is proved to be stable for the case of linear constraints, for non-zero regularization and stabilization parameters. For fixed stabilization value, the regularization can be made arbitrarily small, corresponding to arbitrarily stiff penalty forces. The “relaxed” constraint formulation
permits a separation of time scales so that stiff forces are treated as relaxed constraints. Constraint
stabilization makes the stiff forces modeled this way strictly dissipative, and thus, the stepper essentially filters out the high oscillations, but is rigorously symplectic for the rest of the motion. Spook
solves a single linear system per time step and is insensitive to constraint degeneracies for non-zero
2
regularization. In addition, it keeps the constraint violations within bounds of O ( h ) , where h is the
time step. Because it is derived from the discrete variational principle, the stepping scheme globally
preserves the symmetries of the physical system. The combination of these features make Spook a
very good choice for interactive simulations. Numerical experiments on simple multibody systems
are presented to demonstrate the performance and stability properties.
16 :25 – 16 :50 Tuesday
Adaptive numerical integration of complex multibody problems
with application to rolling bearing simulations
Klas Modin*, Dag Fritzson*, Claus Führer**
*SKF Engineering Research Centre, MDC RKS–2
SE–41550 G¨oteborg, Sweden
e-mails: klas.modin@na.lu.se, Dag.Fritzson@skf.com
**Numerical Analysis, Centre for Mathematical Sciences
Lund University, SE-22100 Lund, Sweden
e-mail: claus@maths.lth.se
Large scale industrial multibody problems require adaptive numerical integrators. Classically, the
step size control algorithms in adaptive numerical integrators (e.g. CVODE, DASPK [3]) are based
on assumed continuity properties of general first order ODE:s. Indeed, the standard approach is to
compute local error estimates from past information, and to use a control system for the step size,
with the aim of keeping the local error at each step constant. Such error estimates are derived from
Taylor expansions, typically by comparing two different schemes (e.g. a predictor of low order is
compared with a corrector of higher order). For multibody problems with contacts the continuity assumptions are not always valid, leading to unsatisfatory step size selection. Standard adaptive stepsize control is based on error estimates computed from past information only. Newly it has been
shown that the step-size may instead be controlled instantly [2]. Our approach is to connect instant
physical quantities (e.g. contact power) to the error analysis, and hence also to the step-size control.
We then incorporate the step-size controller with the ODE. This reflects the fact that adaptive control
contributes to the dynamics in an integration process. The ideas are tested in BEAST [1] – an environment for detailed multibody contact simulations developed at SKF. Hence, we aim for realistic
large-scale industrial applications.
7
16 :50 – 17 :15 Tuesday
A Mixed Test Multi Body Modelling Approach to Load Simulation on Aircraft Components
Antonio Vecchio*, Marco Scionti **, Hans Van den Wijngaert *, Matteo Palmonella *
*LMS International
Interleuvenlaan 68, B-3001 Leuven, Belgium
e-mail: antonio.vecchio@lms.be , web page: http://www.lmsintl.com
**University of Catania
D.I.I.M., Viale Andrea Doria 6, 95125 Catania, Italy
In aerospace industry safety is a crucial element leading aircraft makers to accurately design
critical structural components. Most of the aircraft components are subjected to periodic visual
inspection that allows early detection of damage and taking the required maintenance actions. In
the quest for more advanced methods for structural health monitoring and damage tolerance assessment, aerospace industry has introduced the safe-life and the fail-safe design approach. In order to
perform the safe-life and fail-safe analyses load-spectra and stress-spectra acting on the component under
analysis must be defined.
Structural load analysis consists of the determination of the loads acting on the aircraft structure
for flight maneuvers, flight in turbulence, landing and ground-handling condition. Methods in use to
determine those loads can be quite complex as they deal with the total aircraft solved as complete
system. In such methods, loads are distributed over all the most important structural components under form of panel aerodynamic and inertia loads. The resulting system of equations includes the effects of structural dynamics on the airplane response and requires solution of a multi-degree of freedom system.
The large number of different flight conditions that needs to be investigated makes the validation
of results a time-consuming task. Experimental testing may provide a direct way to identify loads
acting on a specific structural component. For small components, this consists of measuring structural stresses resulting from static loads applied to the structure. However, in real flight conditions
critical components may undergo non-proportional forces, which will generate time dependent variation of local stress tensors at any location on the component surface.
In this research a multi-body modelling approach is applied to the wing slat of an Airbus A320 to
simulate load cases on the slat-track. A large amount of in-flight test data were collected during a
test flight on an Airbus A320 and made available for a dedicated screening of load cases that allowed identifying the most critical operating load conditions. Data were acquired during a long period covering cruise flight conditions as well as intensive maneuvers testing.
Next to dynamic and vibrations test data acquired in several locations along the wingspan, a subset of flight parameters were extracted from the aircraft avionics, the corresponding time signals
were used for a correlation analysis between structural dynamics and flight parameters as altitude,
angle of attack, slat position, etc. This allows isolating a subset of critical load cases for a numerical
simulation of the damage distribution on the surface of the slat track. In parallel, a multi-body dynamics model is used to simulate some additional critical load case as for instance non-symmetrical
loading of the slat, load unbalance generated by a failure in the activation of one of the slat tracks,
etc.
This virtual simulation allows including in the successive fatigue analysis load cases that are unusual in testing conditions, but that can occur in real working conditions. Comparison between load
cases as extracted from dynamic in-flight data and the numerical computation performed on the
multi-body model shows that numerical multi-body simulation is a valid complement for load analysis as it allows reducing testing time while offering a large possibility for virtual simulation of additional load conditions
8
17 :15 – 17 :40 Tuesday
Multibody Modelling and Dynamical Analysis of the worm gear
drive system
Rostyslav Stolyarchuk
State S&R Institute of Information Infrastructure, Lviv, Ukraine
rostyks@yahoo.com
The presented work is currently focusing on dynamic analysis of a worm gear drive mechanical system by multibody contact simulation of the varying force constraint problem and efficiency analysis.
Such systems can be modelled by a Two-Sided Wedge Mechanism (TSWM )as Designed Dynamical Model where worm gearmesh friction contact is exhibited on a plane model by two slope active
contact lines between conjugated wedge-like rigid and inertial bodies via the presence of backlash.
This model has been developed to simplify visualization of internal contact conditions and
force/reactions distributions resembling a screw-like worm gear mesh. The different dynamical regimes of motion (called as “tractive” and “inverse-tractive”) are resulted in varying internal force
distribution with respect to the type of a Force Transfer Function between reduced dynamic reactions
and appropriate force constraints. The transition from one regime to another physically deals with
instantaneous loose and re-establishing sliding friction contact due to the backlash and resulted in
discontinuous events at the force contour/flow level. The mathematical problem formulation results
in residual ODE form with unpredictable discontinuities in the right-hand side within two types of
the Force Transfer Functions. An additional Switching function for control numerical realization or
motion in different regimes has been included. The problem has been tested by MATLAB ODE
solver with added switching conditions incorporated directly into continuous part of ODE system.
The reduced modal parameters for a real-life Motor Operated Valve with worm gearbox has been
used in acceptable range for the simulation purpose.
9
19 :00 – 22 :00 Tuesday
CONFERENCE DINER
Invited Speaker Sten Henriksson - Fifty years with computers.
A talk with a mixture of personal recollections and reflexions on the intellectual contributions of
computers to science. Sten Henriksson is a semiretired senior lecturer at the Computer Science department of Lund University. Starting as a programmer with the SMIL computer, he got his Ph.D. in
numerical analysis (approximation theory). His interest later turned to human-computer interaction.
10
Wednesday 31st of October
11
13:10 – 14:00 Wednesday
KEYNOTE: Model-Based Design at an Enterprise Level:
New directions in modeling and simulation research
Pieter J. Mosterman
Senior Research Scientist
The MathWorks, Inc.
Natick, MA 01760
USA
The analysis and design of engineered systems has a long relied on conceptual models, typically of a
computational nature. Increasing complexity of systems under design as witnessed over the past two
decades has been the driver to support modeling in two directions: (i) provide more sophisticated
modeling languages and (ii) facilitate automation by means of computer aided design tools. ModelBased Design exploits the use of models in many facets of system design; as an executable specification derived from requirements, during early design stages, in prototyping in various forms, for detailed component-level and system-level design, for production code generation, to perform calibration and test design, etc. Each of these usages pose different requirements on the modeling languages
and tool support. In addition, enterprise-wide use of modeling has presented a need to combine, integrate, and reconcile models of different views on the same system under design. The resulting use of
modeling and simulation has put forward a new set of research challenges. In particular, tool infrastructure has become critical to the design of modern engineered systems. This presentation provides
an overview of tool infrastructure that is essential for the successful adoption of Model-Based Design at an enterprise-wide level. The resulting challenges are argued to be important drivers in directing research in the field of modeling and simulation to help elevate the industrial design process to
the next level.
Pieter J. Mosterman is a senior research scientist at The MathWorks, Inc. in Natick, MA. Before, he
was a research associate at the German Aerospace Center (DLR) in Oberpfaffenhofen. He has a
Ph.D. degree in Electrical and Computer Engineering from Vanderbilt University in Nashville, TN,
and a M.Sc. degree in Electrical Engineering from the University of Twente, Netherlands. His primary research interests are in Computer Automated Multiparadigm Modeling (CAMPAM) with
principal applications in training systems and fault detection, isolation, and reconfiguration. For
this, he designed several modeling and simulation environments such as the Electronics Laboratory
Simulator, which was nominated for The Computerworld Smithsonian Award by Microsoft Corporation, and HYBRSIM. He was awarded the IMechE Donald Julius Groen Prize for his paper
"HYBRSIM-A Modeling and Simulation Environment for Hybrid Bond Graphs".
Dr. Mosterman is currently Editor-in-Chief of Simulation: Transactions of The Society for Modeling
and Simulation International for the Methodology section, and Associate Editor of IEEE Transactions on Control System Technology and of Applied Intelligence. He was Mechatronics Area Editor
of Simulation: Transactions of The Society for Modeling and Simulation International and Guest
Editor of special issues of ACM Transactions on Modeling and Computer Simulation and IEEE
Transactions on Control Systems Technology on the topic of CAMPAM. Dr. Mosterman co-chaired
the 14th International Workshop on Principles of Diagnosis (2003) and the annual International
Bellairs CAMPaM Workshop since 2004. He was the Invited Sessions Chair of the 2006 IEEE International Symposium on Computer Aided Control System Design, and Program Chair of the 2007 International Conference on High Level Simulation Languages and Applications as well as the track
on Model-Based Design for Embedded Systems at the 2007 and 2008 Design Automation and Test in
Europe Conference and of the track on Computational Modeling and Simulation of Embedded Systems at the 2007 Summer Computer Simulation Conference.
12
SESSION 3.1:
Session Chair: Bernt Lie
10:00 – 10:25 Wednesday
Simulation and optimization of high power super heater reflectors usable in electrical furnaces for heat loss reduction.
Kourosh Mousavi *, Takami1 Örjan Danielson **, Jafar Mahmoudi**
*KANTHAL researcher and Ph.D. student in Mälardalen University,
**Ph.D. in KANTHAL
**Professor in Mälardalen university
Kourosh.mousavi.takami@mdh.se, örjan.danielson@kanthal.com, jafar.mahmoudi@mdh.se
Super heater elements concentrated in very high and clean power at temperature up to 1800 ºc is
needed for aluminium melting, oil drying equipment using for power transformers and same that material. There fore, always it is need to minimize power consumption and increase effectiveness at
power reflector customers. This means to minimize heat losses in the High Power Reflector, and to
optimize heat performance with respect to output power and temperature distribution. This can be
done by altering the size and form of the insulating ceramic fiber to have as much power as possible
'reflected' out from the module. It can also be done by altering the size and form of the element.It has
been done and the task solved mainly by heat transfer simulations, using COMSOL and MTLAB
software’s. To minimize the number of different designs (and thus the number of calculations) were
studied; it is also done that statistical methods for experimental design used to determine which designs should be used in the calculations.When an optimal solution has been found, it has been tested
at Kanthal's facilities, with two reference designs.
10:25 – 10:50 Wednesday
Aspen HYSYS Simulation of CO2 Removal by Amine Absorption
from a Gas Based Power Plant
Lars Erik Øi
Telemark University College, Norway
lars.oi@hit.no
A simplified combined cycle gas power plant and a MEA (monoethanol amine) based CO2 removal
process have been simulated with the process simulation tool Aspen HYSYS. The thermodynamic
properties are calculated with the Peng Robinson and Amines Property Package models which are
available in Aspen HYSYS. The adiabatic efficiencies in compressors, gas turbines and steam turbines have been fitted to achieve a total thermal efficiency of 58% in the natural gas based power
plant without CO2 removal. The efficiency is reduced to about 50 % with CO2 removal. The CO2
removal in % and the energy consumption in the CO2 removal plant are calculated as a function of
amine circulation rate, absorption column height, absorption temperature and steam temperature.
With CO2 removal of 85 %, heat consumption is calculated to 3.7 MJ/kg CO2 removed, close to a
literature value of 4.0 MJ/kg CO2
13
10:50 – 11:15 Wednesday
Modelling of Electrical Faults in Induction Machines Using
Modelica
Dietmar Winkler Clemens Gühmann
Technische Universität Berlin
Department of Electronic Measurement and Diagnostic Technology
{Dietmar.Winkler|Clemens.Guehmann}@TU-Berlin.de
This paper presents the modelling of electric faults in induction machines using ModelicaR as modelling language. The primal application for such fault simulations is the ever increasing application of
electric machines in cars. In this paper you will find a short introduction of how to model an electrical machine and their most common failures. The modelling is done in the phases reference frame
rather than in a field-oriented reference frame. The reason for this is given in this paper. The simulation models work and simulation results are given. The verification of the simulation results with a
motor test-bench is still an outstanding task at which we are currently working on.
11:15 – 11:40 Wednesday
Adaptive Reduction of Finite Element Models of Complex Mechanical Components
Mats G Larson and Håkan Jakobsson
Department of Mathematics
Umea University
S-901 87 Umea
Sweden
There is currently an increasing interest in using more detailed finite element models of individual
components in multibody simulations. However, due to the large amount of degrees of freedoms in
such models we need to develop so called reduced models with fewer degrees of freedoms. There
are several classical approaches to reduction, including approximations based on modal analysis and
the Craig-Bampton method where the modes correspond to nodewise deformations at the surface.
The Craig-Bampton approach is efficient when the possible load area contains only a small amount
of degrees of freedom otherwise the reduction is not sufficient. In industrial examples the number of
surface nodes in the load area may often exceed 10k. In this work we will investigate an adaptive
load dependent approach which is based on using modal basis functions to describe the global deformations of the component and Craig-Bampton modes to capture a localized load and to provide
an accurate coupling to the modal basis functions. The Craig-Bampton modes are only activated in a
set containing the support of the surface load and may be changed dynamically, depending on the
load situation, in a simulation. We investigate how the overall accuracy depends on the number of
modal basis functions and the size of the area where the Craig-Bampton modes are activated. We
also derive basic a posteriori error estimates which may be used to construct adaptive algorithms for
automatic tuning of the number of modal basis functions as well as the size of the area where the
Craig-Bampton modes are activated. In our examples we study industrial components including a
cage in a roller bearing.
14
11:40 – 12:05 Wednesday
Reconstruction of a Rollover Ejection Case Using Vehicle Dynamics and Occupant Simulation with Human Body Models
Brown, M. Jones, M. Mahangare, M.
Advanced Simtech
www.advancedsimtech.com
Accident reconstruction aims to represent the physics of vehicle driving and impact, as well as human driving and impact biomechanics, using manageable models with predictive capabilities. While
vehicle design is increasingly Computer Aided Engineering (CAE) driven, the advanced modelling
methods developed for CAE are not necessarily straightforward to employ in accident reconstruction. CAE models are often dedicated to specific (laboratory) test conditions for pre-impact driving
or purely for impact, while accident reconstruction often addresses complex real world cases requiring integrated simulation of pre-crash and crash events. Furthermore vehicle and restraint models
developed for design may not be accessible to accident investigators.
This paper addresses the process of building and utilising detailed MADYMO computer models for
accident reconstruction. Key physical components include: tyre to road or soil interaction, suspension, structural crash, restraints, while the human factors include: driving behaviour, and injury biomechanics.
The reconstruction process is detailed for a rollover ejection case. The reconstruction was conducted
in two separate phases. The first phase was to obtain an accurate motion for the vehicle during the
event, using a vehicle dynamics reconstruction simulation that was correlated to the physical evidence and tyre markings left at the scene. The second phase was to understand the opportunity of
ejection using an occupant kinematics simulation.
15
SESSION 3.2:
Session Chair: Jafar Mahmoudi
10:00 – 10:25 Wednesday
Modeling and simulation of a multi phase semi-batch reactor
Anna Nyström
Mathematical Sciences, Chalmers University of Technology and Mathematical Sciences, Göteborg
University
anno@chalmers.se
The operation of an industrial semi-batch reactor, in which the bulk chemical EHEC, ethyl hydroxyethyl cellulose, is produced, is studied and simulated. In the reactor a strongly exothermic polymerization reaction takes place followed by a slightly exothermic reaction, and we want to minimize the duration of the operation of the process. Various operational as well as quality and safety
related constraints have to be met during the batch. The complete process model, derived from
measurements, first principles, and reasoning about effects on molecular level, is stated. The model
includes heat and mass balances of the reactor, a pressure model, models of PID controllers, the
jacket and the condenser. Technical limitations, for instance maximal and minimal jacket temperature changes due to limitations in the heat exchanger, have been modeled as constraints.
The equations have been implemented in SIMULINK, MATLAB and the model predicts the
process variablesrather well over time. During the first reaction, the model is not able to reproduce
the jacket temperature to the desired accuracy, but the other variables have acceptable predictions.
An optimization problem is formulated, wherein the total batch time is minimized under the constraints of the differential algebraic equation system and other constraints originating from the process, for instance limited pump capabilities.
As a first step in optimizing the operation of the process, a series of simulations has been performed in order to decrease the total batch time. It is concluded that a 10 % shorter batch time than
today is possible if the quality is discarded, and a 5 % shorter batch time can be reached while using
the existing requirements for the quality.
10:25 – 10:50 Wednesday
Eclipse Support for Design and Requirements Engineering
Based on ModelicaML
Adrian Pop, Vasile Băluţă, Peter Fritzson
Programming Environments Lab, Department of Computer and Information Science
Linköping University, SE-581 83 Linköping, Sweden
{adrpo,x07vasba,petfr}@ida.liu.se
In order to support the development of complex products, modeling tools and processes need to support codesign of software and hardware in an integrated way. Modelica is the major object-oriented
mathematical modeling language for component-oriented modelling of complex physical systems
and UML is the dominant graphical modeling notation for software. The ModelicaML UML profile
integrates Modelica and UML to support engineering of whole products. In this paper we present the
Eclipse ModelicaML implementation and integration with the MDT Eclipse plugin, with emphasis
on requirements support.
16
10:50 – 11:15 Wednesday
Intelligent modelling of a fluidised bed granulator used in production of pharmaceuticals
Esko K. Juuso
Control Engineering Laboratory, Department of Process and Environmental Engineering, P.O.Box
4300, FI-90014 University of Oulu, Finland
esko.juuso@oulu.fi
The aim of dynamic modelling and simulation is to improve the control of the fluidised bed granulator. Modelling and simulation was done on the basis of data collected from several test campaigns.
Several modelling methodologies have been compared in Matlab-Simulink environment. A solution
based on dynamic linguistic equation models was chosen. The main input variables are humidity difference between incoming and outgoing air, temperature difference between inflowing air and granule and the rate of inflowing air. The final output is the estimated granule size but the overall models
contains also dynamic models for temperature and humidity. The simulator combines several models
which are specific to the operating conditions. According to the results, the spraying and drying
processes included short-duration periods. Extension to fuzzy LE models provides useful information about uncertainties of the forecasted granulation results. The complexity of the models is increased only slightly with the new system based on the extension principle and fuzzy interval analysis.
11:15 – 11:40 Wednesday
A Meta-Modeling Environment for Mechanical System CoSimulations
Alexander Siemers
Dept. of Computer and Information Science
Linköpings universitet, SE-581 83 Link¨oping - Sweden
alesi@ida.liu.se
Dag Fritzson
SKF Engineering Research Centre MDC, RKs-2
SE-415 50 Göteborg, Sweden
Dag.Fritzson@SKF.com
A general approach for modeling of mechanical system co-simulations is presented that is built upon
the previously defined general framework for TLM cosimulations and co-simulation meta-modeling
[1]. Co-simulation is one technique for coupling different simulators into one coherent simulation.
Existing cosimulation applications are often capable of interconnecting two specific simulators
where a unique interface between these tools is defined. However, a more general solution is needed
to make co-simulation modeling applicable for a wider range of tools. Any such solution must also
be numerical stable and easy to use to be applicable by a larger group of people.
In this work the concept of meta-modeling is applied to mechanical co-simulation. Several toolspecific simulation models can be integrated and connected by means of a meta-model, where the
meta-model defines the physical interconnections of these models.
A general meta-modeling process is described that represents the basis for this work. A metamodeling language (MML) has been defined to support the modelling process and store the metamodel structure. Besides elements for physical interconnections, etc., the language also defines
graphical elements that can be used for meta-model visualization. All proposed solutions are general
and simulation tool independent.
A fully functional modeling environment has been created to make meta-modeling applicable.
The modelling environment supports easy encapsulation and integration of simulation tool-specific
models. Each simulation tool implements a single, well defined co-simulation interface. All inter-
17
faces implement a numerically stable method for force/moment interaction. The presented environment features a graphical user interface for cosimulation modeling with support for three dimensional visual representation of the co-simulation model including all its components.
SESSION 4.1:
Session Chair: Bipin Patel
13:00 – 13:25 Wednesday
State/Parameter Estimation of a Small-scale CHP model
Juan I. Videla, Bernt Lie
Telemark University College
juan.videla@hit.no, bernt.lie@hit.no
The state/parameter estimation problem is studied for a small-scale ICE CHP model. Three main
groups of estimators with signi.cant performance and complexity di¤erences are analyzed: the Extended Kalman Filter (EKF) as an extension of the classical Kalman filter, the generalized unscented
Kalman filter (UKF) that uses the unscented transformation, and particle filtering like the particle
fillter with resampling (PFr) and the Ensemble Kalman Filter (EnKF)
The internal combustion engine is modeled as a mean-value engine model connected with a static
generator model and the heat recovery circuit is modeled with two lumped heat exchanger models,
one for the coolant circuit and the other for the exhaust gases. The coolant circuit is connected with
the engine through a lumped inner engine thermal model. Experimental data sets are arti.cially generated to test the different estimators. Dynamic parameters of the mean-value engine model are identify when the CHP model is simulated in open loop. Additionally, relevant heat transfer coefficients
of the heat recovery circuit are monitored when the model is simulated in closed loop.
13:25 – 13:50 Wednesday
Using an adaptive FEM to determine the optimal control of a
vehicle during a collision avoidance manoeuvre
Karin Kraft and Stig Larsson
Mathematical Sciences, Chalmers University of Technology and Göteborg University
karin.kraft@chalmers.se
stig@chalmers.se
Mathias Lidberg
Applied Mechanics, Chalmers University of Technology
mathias.lidberg@chalmers.se
The optimal manoeuvering of a vehicle during a collision avoidance manoeuvre is investigated. A
simple model where the vehicle is modelled as point mass and the mathematical formulation of the
optimal manoeuvre are presented. The resulting two-point boundary problem is solved by an adaptive finite element method and the theory behind this method is described.
18
13:50 – 14:15 Wednesday
Towards Run-time Debugging of Equation-based
Object-oriented Languages
Adrian Pop and Peter Fritzson
Programming Environments Laboratory, Department of Computer and Information Science,
Linköping University
{adrpo, petfr}@ida.liu.se
The development of today’s complex products requires advanced integrated environments and modeling languages for modeling and simulation. Equation-based object-oriented declarative (EOO) languages are emerging as the key approach to physical system modelling and simulation. The increased ease of use, the high abstraction and the expressivity of EOO languages are very attractive
properties. However, these attractive properties come with the drawback that programming and
modeling errors are often hard to find. In this paper we propose an integrated framework for runtime debugging of equation-based modeling languages. The framework integrates classical debugging techniques with special techniques for debugging EOO languages and is based on graph visualization and interaction. The debugging framework targets the Modelica language.
14:15 – 14:40 Wednesday
Dynamic Model of a Bubbling Fluidized Bed Boiler
Tuomas Kataja, Yrjö Majanne
Tampere University of Technology, Institute of Automation and Control
P.O. Box 692, FI-33101 TAMPERE
tuomas.kataja@tut.f, yrjo.majanne@tut.fi
A dynamic model for a high-volatile solid fuel fired bubbling fluidized bed boiler is presented. The
model consist of an air-flue gas model which includes a furnace model describing combustion in a
bubbling fluidized bed and a model for a water-steam circuit describing heat transfer from hot flue
gases to water and steam. The versatile furnace model takes account of quality parameters of fuel so
that the effects of moisture, particle size, heat value, and the amount of volatiles can be simulated.
The model is based on the first principles mass, energy, and momentum balances. Results from validation of the model against a bubbling fluidized bed boiler process data are presented. The validation
showed that the model can describe the dynamics and static gains of the process very well.
19
SESSION 4.2:
Session Chair: Erik Dahlquist
13:00 – 13:25 Wednesday
Simulation of energy in the building and design a new intelligent building with controllable and wise devices
Kourosh Mousavi, S.M.Esmail Mousavi Takami, Jafar Mahmoudi
Kourosh.mousavi takami@mdh.se, architecttakami@yahoo.com, jafar.mahmoudi@mdh.se
Building energy simulation is important for the study of energy efficiency in buildings. An energy
saving control system of lightning, heating and variable-air-volume air conditioner in intelligence
building is simulated in this paper. It reaches good control effect and energy efficiency by making
the best of the advantages of intelligence building. In the regulating period, the lightning lux, heating
flux and air volume is decided by means of feed forward control. The previous turning off period is
determined in the way of penalty function. It has been used in a sample building for developing
building energy standards and analysing energy consumption and conservation measures of buildings.
13:25 –13:50 Wednesday
Identification of a best thermal formula and model for oil and
winding of power transformers using prediction methods
Kourosh Mousavi, S.M.Esmail Mousavi Takami, Jafar Mahmoudi
Mälardalen University, Västerås, sweden
Kourosh.mousavi.takami@mdh.se jafar.mahmoudi@mdh.se
System identification is about building models from data. A data set is characterized by several
pieces of information: The input and output signals, the sampling interval, the variable names and
units, etc. Similarly, the estimated models contain information of different kinds, estimated parameters, their covariance matrices, and model structure and so on. In this paper we collected Temperature of oil and winding in 230/63kv transformer of SARI Substation and considered the winding
temperature for input in the model and oil temperature for out put. After that calculated their data by
MATLAB software and get a new model with the good best fit for the heat transfer from core and
winding to oil. For verification of were calculated results, has been simulated the process in COMSOL Software.
20
13:50 – 14:15 Wednesday
Stochastic simulation of insurance business
Dmitrii Silvestrov, Anatoliy Malyarenko, Evelina Silvestrova
Mälardalen University, Västerås, Sweden
dmitrii.silvestrov@mdh.se, anatoliy.malyarenko@mdh.se, evelina.silvestrova@mdh.se
We present a model for stochastic modeling of insurance business with dynamical control of investments and a pilot program system SMIB (Stochastic Model-ing of Insurance Business).
In our model, we present the business of an insurance company as a multi-component dynamical
system in-cluding basic risk processes, return of investments in-dexes for various types of assets, inflation and other in-formative components.
The project was inspired by activities of the Finish Insurance Modeling Group. This group elaborated and effectively used the global simulation programs for modeling the business of insurance
company. The modeling was performed for a time horizon of 20—30 years with a year time step in
the modeling process. These programs were used to find out regulations for admissible limits of investments to risky stock market under the assumption of stationary distribution of in-vestments between various types of assets.
The basic idea of the program SMIB is to involve Monte Carlo method for producing multiple time
scenarios describing functioning of an insurance company and dynamical behavior of its capital. The
capital can be invested into different types of assets. Each type of investment has its own profitability and risk. The in-surance business and the investment process are simulated with the help of the
multi-parameter non-linear dynamical model. Premiums, claims, and profits of dif-ferent types of
investments are described by equations of autoregressive type.
The main new element of the project presented in the present paper is that modeling of insurance
business is performed for the model, which incorporates non-stationary dynamic threshold strategies
for re-distribution of capital between various types of assets. It means that the quotas of the current
capital corresponding to different types of investment and safety loading coefficients in the next time
period are switch-ing in correspondence with the value of company's capital at the end of the present
time period. The switching levels, or thresholds, are functions of time. Thus the process under consideration is essentially non-linear. Analytical methods do not work here, but Monte Carlo simulation does.
There exist two versions of the SMIB program system. The interface and functional capabilities of
both versions are identical. One version is written in Microsoft Visual C++ using the MFC library
and works under MS Windows. Another version is platform independent and is written in Java.
SMIB is based on a model with time horizon 1--5 years with one-month time step in the modeling
process. The first version of the pilot program system for dynamical control of investments for an insurance company based on this model, have been elaborated, and it is presented in the paper. Experimental studies realized so far show very interesting and non-trivial results concerning shapes of
distribution of capital of insurance company at given time horizon in the case of non-linear investment strategies.
21
14:15 – 14:40 Wednesday
Modelling Space Heating Systems Connected to District Heating in Case of Electric Power Failure
Per-Olof Johansson, M Sc, Patrick Ljunggren, M Sc, Janusz Wollerstrand, Ass. professor
Lund University, Department of Energy Sciences, Division for Efficient Energy Systems
Per-Olof.Johansson@vok.lth.se
Recent year’s extensive power failures have put focus on the importance of secure local production
and distribution of energy. Since district heating (DH) is the dominating heating system in Scandinavia it is of great importance to investigate the possibility for buildings connected to DH to receive
heat during an electric power failure. By using computer models, buildings’ heating systems can be
simulated, and by comparing the results with field studies the model can be evaluated. The model
shows good resemblance with the field study. By using a model the influence of different parameterscan be studied. The influence of, e.g., changed primary supply temperature and outdoor temperature
can be studied.
22
Exhibitors
HyperWorks
http://www.altairhyperworks.se
Altair Engineering, Lund, Sweden
The Open Modelica Project
http://www.ida.liu.se/labs/pelab/modelica/OpenModelica.html
Adrian Pop, David Broman, Håkan Lundvall, Peter Fritzson
Department of Computer and Information Science
Linköping University, Sweden
23