SimMechanics Vs SolidWorks COSMOSMotion

[aerog...@googlemail.com]

Hi All,

Am an experienced Simulink user looking to do some simulation of
mechanisms.
I was looking at importing Solidworks drawings into SimMechanics to do
the analysis but then found COSMOSMotion which looks like a better
tool.

Anyone have experience of simulating mechanical systems with either
package?

Thanks,

Mark.

[Philippe Guglielmetti]

CosmosMotion is integrated in SolidWorks, so it is a "purely
mechanics" simulation system, where you can easily define joints,
restraints, forces in 3D, and you can take advantage of Animator and
CosmosWorks features to make animations and FEA studies of critical
parts.
However, when you need to simulate a system with feedback, control,
complex trajectory generation, you'll need to dive into its (very) old
fashioned ADAMS programming language. See the online help about
"adams" to get the picture.

SimMechanics is an extension of a great "system oriented" simulation
package : Simulink. In Simulink your system is modelled as connected
blocks which define "transfer functions" between inputs and output.
Each of your parts and each of your joints will be converted in a
block by the cad-2-simmechanics tool, then you'll complete the diagram
by adding source signals, feedback loops and automation devices in a
much more comfortable manner than coding in adams.

A very interesting alternative to Simulink is Dymola by http://www.dynasim.com/
,which relies on an open-source language called Modelica http://www.modelica.org/.
It'd definitely be my choice for complex systems.

Block diagram oriented simulation is in trouble with contact problems,
for example when a part is in a conveyor and falls on another device.
You'd need to define all the possible contacts as blocks, while 3D
tools such as CosmosMotion use general collision (interference)
detection. However, Block diagrams allow to model contacts and
friction with much higher detail than the "impact" model used in 3D
(CosmosMotion). The combination of the general collision detection
with impact forces and frictionresults in long simulation time and,
quite often, wrong results.

A difficulty with all tools including CosmosMotion is the conversion
of assembly "mates" into cinematic joints (revolute, prismatic...),
while handling "flexible" assemblies and other cad-related stuff. It
works on simple models where mates have been carefuly defined using
gormetry on connected parts. If your model has "design mates" such as
the distance between reference planes of two distant parts, or if you
didn't select the same parts to define the concentricity and the
coincidence of a revolute joint, or didn't block the free motion of
your screws, prepare to spend some time clearing the jungle of the
cinematic joints...

Hope this helps.

( Looks like this answer is long enough to become a post on a blog
soon ;-)

[aerog...@googlemail.com]

On Apr 28, 3:22 am, Philippe Guglielmetti <goo...@goulu.net> wrote:
> CosmosMotionis integrated in SolidWorks, so it is a "purely

> mechanics" simulation system, where you can easily define joints,
> restraints, forces in 3D, and you can take advantage of Animator and
> CosmosWorks features to make animations and FEA studies of critical
> parts.
> However, when you need to simulate a system with feedback, control,
> complex trajectory generation, you'll need to dive into its (very) old
> fashioned ADAMS programming language. See the online help about
> "adams" to get the picture.
>

> SimMechanicsis an extension of a great "system oriented" simulation

> package : Simulink. In Simulink your system is modelled as connected
> blocks which define "transfer functions" between inputs and output.
> Each of your parts and each of your joints will be converted in a

> block by the cad-2-simmechanicstool, then you'll complete the diagram

> by adding source signals, feedback loops and automation devices in a
> much more comfortable manner than coding in adams.
>

> A very interesting alternative to Simulink is Dymola byhttp://www.dynasim.com/
> ,which relies on an open-source language called Modelicahttp://www.modelica.org/.

> It'd definitely be my choice for complex systems.
>
> Block diagram oriented simulation is in trouble with contact problems,
> for example when a part is in a conveyor and falls on another device.
> You'd need to define all the possible contacts as blocks, while 3D

> tools such asCosmosMotionuse general collision (interference)

> detection. However, Block diagrams allow to model contacts and
> friction with much higher detail than the "impact" model used in 3D
> (CosmosMotion). The combination of the general collision detection
> with impact forces and frictionresults in long simulation time and,
> quite often, wrong results.
>

> A difficulty with all tools includingCosmosMotionis the conversion

> of assembly "mates" into cinematic joints (revolute, prismatic...),
> while handling "flexible" assemblies and other cad-related stuff. It
> works on simple models where mates have been carefuly defined using
> gormetry on connected parts. If your model has "design mates" such as
> the distance between reference planes of two distant parts, or if you
> didn't select the same parts to define the concentricity and the
> coincidence of a revolute joint, or didn't block the free motion of
> your screws, prepare to spend some time clearing the jungle of the
> cinematic joints...
>
> Hope this helps.
>
> ( Looks like this answer is long enough to become a post on a blog
> soon ;-)

Hi Philippe,

Thanks for your comprehensive reply!

All your points are noted and you perfectly describe the problem I
hope to solve.
My system is a complex conveyer, lots of small objects pass through. I
want to model the path of these objects through the system which
includes elevators, rotating, linear transports and gravity drops from
one path to another.

And again you're right, the system consists of main closed loop
controllers for the drive mechanisms and sensor feedback so Simulink +
Simmechanics looks like the winner so far for its flexibility and
capacity for modeling the state machine as well.

It's always good to here the thoughts of people that have used the
tools in real applications and I thank you for yours.

Best regards,

Mark.