Archived OpenModelica forums. Posting is disabled.

Alternative forums include GitHub discussions or StackOverflow (make sure to read the Stack Overflow rules; you need to have well-formed questions)


Forgot password? | Forgot username? | Register
  • Index
  • » Users
  • » tklinge
  • » Profile

Posts

Posts

Hello, at the moment I do a research about the features of openmodelica and it's usability to create simulation models, that can be used in an own c++ realtime application.

The actual state of my research is:

1) I made a test model in openmodelica which works fine. I use the dassl solver.
2) I export the model as FMU 2.0 (ME&Co).
3) I wrote a class encapsulating the fmi-library calls. This class I use in a standalone realtime simulation. I can load the FMU and call fmi2_do_step.

Until here, I run into trouble: After some investigations (because I am new to openmodelica) I have found that the co-simulation only supports forward euler. But because of the model I need dassl algorithm as solver (best: same solver like openmodelica uses).

How to proceed?

1.) Is it planned to implement dassl in FMU cosimulation in the near future (< 3 month)?
2.) Should I implement an own dassl solver ? But I am new to this and have only small knowledge on solvers. Does anybody do so already and can give some hints / code segments?
3.) Is it possible, to "extract" the dassl solver from openmodelica and link it to an own standalone application?
4.) Should I kick the FMU concept and should I concentrate on the generated code and integrate it as source + own dassl?
5.) Does someone have any other suggestion? I heard about UPC simulation? What is this, how to do it?

I personally would prefer version 1). During testing with openmodelica (different own model creations) I have seen, that if the dassl solver fails during simulation, automatically a fallback solver is used and the simulation continues. For usage in a realtime system this is what I want!

Regards

Ok, thank you, think you are right. At the moment I checked the 32 bit version of the latest master and it is working. This week I will also check the linux64 bit version.

I use Version v1.12.0-dev, build on suse Linux 64-bit. I programmed a model and want to use it as FMU.
Simulation works, but during simulation I get warnings from solver:
Failed to solve linear System of equations ...
The Default linear Server Fails, the fallback solver with total pivoting is started...

But nevertheless the Simulation works.

Then I export the solver as FMU 2.0 with Modeldescription and Co-Simulation.

When I want to use the FMU there are dependencies to

libModelicaStanardTables.so.0
libModelicaMatIO.so.0
libzlib.so.0

But finally, I can't get it to run in my application, because the Symbol debugString is missing. (I load the FMU using the fmi-library)
It seams, that there is a dependency not covered by the FMU package. I investigated the source and found, that in the file nonLinearSolverHomotopy.h/.c this function is defined.

How to solve this Problem?

Regards


  • Index
  • » Users
  • » tklinge
  • » Profile
You are here: