I am running a system with a winch, hydromotor, pump, directional valve and control valve.
Directional and control valves are supercomponents.
The system is running well since yesterday but it is very low.
I assume the calculation is running on only one core, and I have seven othe cores which are free.
Is there a way to speedup the calculation?
I saw master/slave topic, but no idea where are those related components (I understand we could divide the system into subdomains, and each subdomain could be treated from a core)
Solved! Go to Solution.
there are several ways to try and speed up a model.
First about the usage of multi-core.
When using the hydraulic library.
For any library.
Some general tips.
I hope this helps.
Generic cosimulation technique can be safely used when you can naturally decouple several parts of a system. It means that the 2 subsystems should have a limited (or slow) interaction. Otherwise, in order to correctly capture the interaction between the subsystems, you'll have to use a very small 'sample time' in the cosimulation block parameter list. In that case you may not see a speedup. With pure hydraulic systems it can be challenging to find such a natural split between slow-interacting subsystems!
A demo model ClutchFading illustrates a situation where a split is easily achieved. Thermal effect on one side (slow dynamics) are coupled to mechanical model with higher dynamics, it's ok to create a coupling using cosimulation blocks.
Check the demo, launch simulation from the slave model then from the master.
You may also want to look for additional ways to reduce complexity of your model. For example, if you have multiple small volumes at different places in the hydraulic circuit, we usually recommend creating an equivalent larger volume to reduce the number of state equations that the solver generates when compiling. You can also choose simpler sub-models if the fidelity and accuracy are still acceptable for the quality of the analysis you are looking for. Generally, you want the number of explicit states generated (shown when the compiling window pops up) to be as low as you can get while maintaining accuracy/fidelity. Choosing a faster compiler and splitting the simulation up onto multiple cores for cosimulation are also options, but generally the last option once you have exhausted all other methods for model reduction. I hope this helps as well!
Production: Amesim 14.1, Autonomie R12
1st click, I always recommend:
Check the column Controlled. This will automatically highlight you the biggest "consumer" in your system (interactions between submodel and solver).
A double-click on the row will label you the corresponding element on your sketch.
Now, you know where to start first with your simplification / speed up modifications:
Check the complexity (submodel) of this element; parameter setting; initial conditions etc.
Let me/us know your progress.
Happy modeling ; )