Exploring the Possibilities of Co-Simulation with

CATIA V6 Dynamic Behavior Modeling

Peter Smolek1 , Bernhard Heinzl2 , Horst Ecker1 , Felix Breitenecker3
1
Vienna University of Technology, Institute of Mechanics and Mechatronics
2
dwh simulation services, Vienna, Austria
3
Vienna University of Technology, Institute for Analysis and Scientific Computing
peter.smolek@tuwien.ac.at

Abstract model descriptions and multiple numerical algo-

This study shows how to utilize the CATIA V6 Dy- rithms can be used to study heterogeneous sys-
namic Behavior Modeling (DBM) software envi- tems.
ronment for the purpose of co-simulation of physi- With the CATIA V6 release, an environment
cal models. The implementation of a co-simulation for systems simulation has been implemented – the
using Building Controls Virtual Test Bed (BCVTB) Systems Dynamic Behavior Modeling (DBM) [1] –
is demonstrated and the pros and cons are dis- which in itself consists of an altered Dymola build.
cussed. Further the methods for FMI export and Therefore the CATIA DBM also uses the Model-
import in CATIA are explored with respect to im- ica standard [2], and the subsequent libraries. The
plementing a co-simulation either in CATIA itself layout has been adjusted to fit the general CA-
or other host programs. The two approaches are TIA logic, the functionality of the Dymola soft-
displayed by implementing simple examples. Ulti- ware remains for the most part unaltered. The
mately possible applications for an advanced tool main difference with respect to Dymola structure
to link 3D geometric data and systems simulation, is that the DBM is integrated in the CATIA Prod-
with the potential to perform co-simulation, are uct Lifecycle Management (PLM) solution, there-
presented. fore the data is no longer stored locally but on a
server. On this remote computer the files of mul-
tiple users are stored and the license management
1 Introduction
is performed. As a result, libraries need to be im-
The approach of using system simulations results ported and propagated to the server before use.
in new opportunities for integrated virtual prod- Modelica files can be imported as well as exported,
uct development. Putting the product itself in however the common approach is to work with the
the center of focus and building the different tools DBM projects through the server.
around it allows connecting the different disciplines As part of the CATIA PLM, combining physical
of development, such as Computer Aided Design data with the functionality of the model is real-
(CAD), Computer Aided Engineering (CAE) and ized through the Requirements, Functional, Logi-
system simulation. Co-simulation describes the cal and Physical (RFLP) environment, which can
process of using multiple programs working on the be accessed through the Functional and Logical
same simulation, sharing data with each other. Workbench seen in figure 1. It is possible to de-
This grants the flexibility to use specialized soft- fine logical references, which interact through con-
ware to describe specific parts of the model, for nections and can be arranged hierarchically. The
example analytical, data-based, static or dynamic references in turn can be given a DBM model. At
components. This enables to examine complex the top of figure 1 an example of a logical hierarchy
problems like multidomain simulations, stiff sys- is shown. The small icon in the lower right cor-
tems and many more. In addition it allows the ner of the logical references indicates that a DBM
developers to use the software tool best suited model is implemented. Further, a geometric rep-
for their field of study, collaborating simultane- resentation can be linked to the logical reference
ously on the same model. Thereby using different to represent the physical properties of the logical

ERK'2013, Portorož, A:177-180 177

 guage. Therefore further control over the CATIA
instance is possible, especially remote execution of
simulation runs is practical.

Figure 2: Model in BCVTB; the CATIA simulator

block is responsible for the communication with CA-
TIA.
Figure 1: CATIA Functional and Logical Workbench
with logical references containing a DBM model and To demonstrate the basic data exchange, a sim-
a 3DRepresentation. ple model was implemented consisting of a ramp
signal, defined in the BCVTB environment. The
data signal is then sent to the simulator actor
component, as seen at the bottom of figure 1. As block, which implements the BSD socket commu-
it stands, it is not possible to use general 3DParts nication with the CATIA DBM environment, as
in the RFLP environment but special 3DRepre- seen in figure 2. The DBM model simply consists
sentations. This however is announced to be re- of a gain element, which gets the data value from
solved in future releases. The whole simulation BCVTB, multiplies it by two and returns the re-
can then be run from the Functional and Logical sult to the backbone. The model is shown in figure
Workbench where all the underlying DBM models 3, figure 4 is demonstrating an example simulation
are executed and data is exchanged according to run and depicts the input and output signal (timed
the logical connections. plotter in figure 2). The red graph represents the
2 Co-simulation with BCVTB ramp output, the blue is the returned data of CA-
TIA, which is zero per default during the initial
To prepare CATIA DBM for co-simulation, a soft- data exchange. This simple example is solely sup-
ware solution that is compatible with Dymola is
chosen. The Building Controls Virtual Test Bed
(BCVTB) [3] delivers that in combination with
an interface provided in the Modelica Buildings
Library [4]. The BCVTB was originally used for
buildings simulation, but the ability to link to the
general purpose tools Dymola and MATLAB en-
ables the use in other domains. Since this is a
Modelica library, importing it into the DBM en-
vironment is easily achieved. Models can be built
very similarly as in Dymola, since the function-
ality of the two is basically the same. BCVTB
controls the communication between the backbone
and CATIA and relies on BSD sockets [5] for run-
time data exchange. To minimize startup time,
an existing CATIA instance is accessed. The run-
ning instance can be contacted through a COM Figure 3: Modelica model implemented in CATIA
server, which CATIA already runs natively. This DBM with the BCVTB Modelica block; the data
can be achieved via a Visual Basic script since CA- is received from BCVTB, amplified and returned to
TIA handles macros internally with the same lan- BCVTB.

178
posed to demonstrate the communication between in the DBM as well as the RFLP environment. For
BCVTB and CATIA, no differential equations are this study, the FMUs to be imported are created
solved and the connection to a second simulation with Dymola 2014. First of, the DBM Workbench
software is missing. offers the same functionality as does Dymola in
its current release. It is possible to import FMUs,
link them via appropriate interfaces to the DBM
model and run the simulation. The export of a
model, as well, works with exceptions unproblem-
atic. The more interesting subject is the FMU
export and import of the Functional and Logical
Workbench. As it stands, there is no method for
exporting models as FMUs. The option that is
available is only for exporting already imported
FMUs again and therefore is not suitable for us-
ing RFLP content externally. The import how-
ever is implemented as an extra method of defin-
ing DBM models for logical references. This is a
Figure 4: Output of the timed plotter of BCVTB; red: convenient way of linking FMUs to logical refer-
the ramp, blue and dotted: simulator output. ences, which can be connected to the rest of the
logical hierarchy for data exchange. The second
With this method of employing co-simulation, possibility of importing a FMU is to import it
coupling models implemented in different simula- into a DBM model. This method is more robust,
tion tools with CATIA is achieved. In order to even though some issues regarding updating an
take full advantage of what system simulation in imported FMU do exist. Both methods grant ac-
combination with a CAD software offers, it is nec- cess to co-simulation in the RFLP environment.
essary to couple BCVTB not with the DBM, but
with the Functional and Logical Workbench. Un-
fortunately, with the current release of CATIA
V6 that could not be achieved. The RFLP en-
vironment cannot execute a simulation success-
fully with the BCVTB block as a part of it, even
though the underlying DBM model in itself can
be run. The problem seems to be, that the linked
C-functions are not called properly. A solution
could not yet be found since CATIA does not al-
low insight into the internal processes. It is to be
expected, that this issue will be resolved in the
future, enabling real advantages for co-simulation
with CATIA.

3 Co-simulation with FMI

Another interesting possibility to realize co-simula-
tion with CATIA V6 is by employing the Func- Figure 5: DBM model of a revolvable cylinder; the
tional Mockup Interface (FMI) [6]. This standard- data from the FMU is received through the input sig-
ized interface allows exporting models as Func- nal port, driving the rotation.
tional Mockup Units (FMU), which in turn can
be imported into other simulation tools. It is pos- For demonstration purposes, a simple setting
sible to either export the model alone – the solver was chosen. In the RFLP environment, two logical
of the host then calculates the imported model – references have been defined (figure 1). One holds
or the solver algorithm can be exported with the a normal DBM model, containing solely the im-
model, therefore allowing a co-simulation with a ported FMU. In the second reference a model of a
modelica host, which is further discussed. damped rotation of a cylinder is implemented, see
CATIA supports the export and import of FMUs figure 5. A torque drives the rotation and receives

179
 sembly Design, since the program already has the
information of the relationship between different
components. Therefore realizing dynamic analysis
of complex components in a short timeframe would
become realistic. Testing a product virtually be-
fore prototypes are built has the benefit, that com-
plex interactions of parts are incorporated in the
development. This results in reduction of devel-
opment time and cost of mechanical parts.

References
[1] Dassault Systemes: CATIA V6R2013X Port-
folio, 2013, http://www.3ds.com/products-
Figure 6: Plotting window of CATIA in the Functional services/catia/portfolio/catia-version-6/v6-
and Logical Workbench; red: the angle of the cylinder portfolio/
over time.
[2] Modelica Association: Modelica Specification,
version 3.3, 2012, modelica.org
its amplitude from the FMU. Using the appropri-
ate block in the DBM model and linking it with [3] M. Wetter: Building Controls Virtual Test Bed
a 3D representation realizes the mechanical arm. User Manual Version 1.3.0, 2013, simulationre-
Figure 1 shows the logical hierarchy and the cylin- search.lbl.gov/bcvtb
der in motion. Figure 6 is an example of a plotter
window showing the angle of the mechanical arm [4] M. Wetter et al.: Modelica Buildings Library,
plotted over time. 2012, Journal of Building Performance Simu-
With the presented methods, setting up a co- lation, (pre-print) 2013
simulation using FMUs in CATIA is possible. How-
[5] M. Wetter: Co-simulation of building energy
ever, the potential is constricted by the existence
and control systems with the Building Controls
of a way to export a model via FMI. It is to hope,
Virtual Test Bed , 2011, Journal of Building
that with further development of the FMI stan-
Performance Simulation, 4(3):185-203, 2011.
dard more simulators will add support for export-
ing FMUs, thus allowing a flexible choice of soft- [6] Modelica Association: Functional Mock-
ware for co-simulation. up Interface for Model Exchange and Co-
Simulation, 2.0 Beta 4, 2012, www.fmi-
4 Outlook standard.org
As the situation currently presents itself, the u-
nique opportunities of combining CATIA V6 with
a system simulator cannot be efficiently utilized
for co-simulation yet. Once the software is fur-
ther developed, the usability will be enhanced and
integration of the physical data will enable effi-
cient knowledge management across various do-
mains. Furthermore, whole new options arise from
the possibility of not only being able to drain in-
formation of CAD data but to alter the geometry
data itself. It is conceivable to build a model of a
mechanical systems simulation inside an optimiza-
tion loop. The optimization, for example a genetic
algorithm, can then adapt the geometry to opti-
mize e.g. the distribution of cross sections of a
shaft.
One can imagine the possibility to automatically
extrapolate a mechanical model from CATIA’s As-

180