Beta2 vs Beta4: worsening performance and not convergence of a simulation

[andreabrambilla82@…]

Hi,

I recently switched from Beta2 to Beta4 for Windows and I observed a considerable worsening of the performance, in particular:

• my simulations run much slower (even 5 times slower);
• OpenModelica crashes quite often when I save the changes to the models and the saving operation is much slower (a few seconds against a few minutes);
• when clicking on a component to display the mask where the parameters can be changed, in Beta2 the mask appears almost instantaneously while in Beta4 the mask appears even after 15 minutes.

Other than this, I have a quite simple model for simulating water-hammer which runs without any problem on Beta2 but it crashes on Beta 4 because the medium (compressible water) reaches a too high pressure. attached you can find the model.
Thanks and best regards,
Andrea.

[andreabrambilla82@…]

water-hammer setup: works with Beta2 but not with Beta4

[Christoph <buchner@…>]

A connected issue (at least for your last bullet point) is #2882.

[Christoph <buchner@…>]

I can report that using r23143, the WaterHammerValve simulation completes, although it does take a very long time. I'm unable to judge the accuracy of the results, there's quite a lot of (numerical?) oscillations in there, possibly that's to be expected?
tail of output log:

LOG_STATS		| info	 | solver: DASSL
|		| |	 | | 569513 steps taken
|		| |	 | | 582812 calls of functionODE
|		| |	 | |  1213 evaluations of jacobian
|		| |	 | |  1015 error test failures
|		| |	 | |     0 convergence test failures
LOG_STATS		| info	 | ### END STATISTICS ###

[anonymous]

Yes, it works with r23143, thank you. It was not working with a previous version, no idea why.
The oscillations are physical, they are due to pressure waves propagation back and forth. It is also reasonable that the computational time is high.
Thank you.
Regards,
Andrea.

[Francesco Casella]

With the latest OpenModelica-v1.13.0-dev-63-g453e1c7-64bit the model simulates in about two minutes, the statistics are similar to those reported in comment:2. I also understand the large number of steps is due to the high-frequency pressure oscillations in the system.