Modelica and discontinuities

[lutz.d....@googlemail.com]

Dear Jean-Philippe,

today I want to send you an example for the maybe not completely satisfying treatment of discontinuities.

The example is a hard-switching boost converter. It includes some self-made models which are more or less close to the Modelica standard models (MSL). The same simulation works flawlessly in Scicoslab.

I would prefer to go on with NSP (and not to migrate to JModelica.org or OpenModelica), since post-processing is much more powerful. Can you please help me?

I suggest, that after a switching event, the BDF algorithm (i.e. the DAE solver) should:
- reduce the order to 1 (=> implicit Euler)
- switch to minimum step width
- ignore the error of the first time step.

I do not know, which DAE solver you use, but the above strategy works satisfying for BDF (and many simulators employ a linear multistep approach).

Thank you very much in advance.


Finest regards,

Lutz

[Jean-Philippe Chancelier]

Dear Lutz,

I am starting to look at your example, but I need some more informations.
If I understand correctly your mail, you seams to say that the model
works fine in scicoslab and is not satisfying when using nsp/scicos. Is it correct ?
Best Regards
jpc

[lutz.d....@googlemail.com]

Dear Jean-Philippe,

I am very sorry for the late answer. I was terribly busy with the lectures ...

Yes, it is precisely as you say: the model works fine in scicoslab but does finish in scilab nsp (breaks after a few microseconds with an initialisation error message).

It would be fine if you could take a look at it.


Finest regards,

Lutz

[jpc]

Dear Lutz,
I am investigating the differences and have not found yet why it does not work in
nsp as in scicoslab as it should. I'll let you know when it will be ok.
Regards
jpc

[lutz.d....@googlemail.com]

Dear Jean-Philippe,

thank you very much. Do not be stressed. I am waiting patiently.
Thank you for your work.

Best regards,

Lutz.