Network coupler for gas

Hello. Does anywhere exist the network coupler for gas? I don’t see it in the simscape library. See the attachment. I need this element to split a gas model on 2 parts to simplify solving on real-time machine. Did anybody come across with this issue? Thanks.

Answer 1

[Yifeng Tang]

Network coupler for Gas or Moist Air domain doesn't exist as of R2024b.

I see two options:

(1) follow the structure of such component for the TL domain, and write your own coupler for the Gas domain. The key component is the "partialChamber". You may compare it to the regular constant volume chamber and understand the type of change needed to be made for the Gas domain.

(2) ask yourself whether the original model has been optimized numerically and whether splitting it is the only path forward. More details, e.g. an attached model, will help clarify on this.

[Alexander]

Hello Yifeng,

Thank you for answer. I understood the first piece of advice.

The model is in the attachment. It's a model of air conditioning system. The system consists of the two symmetrical parts. So it seems good for splitting. The real-time machine doesn't manage with full system.

Could you analyze the solver and help with optimization?

Thank you vary much.

[Yifeng Tang]

Hi @Alexander

I took a look at your model. While I don't fully understand the interactions betwen the two symmetrical sides, it seems to me, with all the check valves in place, the two sides are more or less independent of each other. Is that right? For example, the flow from the top side will never enter the bottom side?

If this is indeed the case, I think it's possible to totally separate the two parts as two models. You may need a copy of your "ЗУ4 (5735)", "ЗРУ5 (5751)", "ЗРУ6 (5751)" so they can both flow/leak to outside.

It's nice that the Simulink global solver ode1be is working OK for your model right now. However, I still would like to you check out the documentation page on running real-time simulation with Simscape.

https://www.mathworks.com/help/releases/R2023b/simscape/real-time-simulation.html

A key part of the workflow is to understand the time-step information when using a variable step solver (e.g. daessc), to infer how the model with behave when using a fixed-step, fixed-iteration/cost solver. A super useful tool is the Solver Profiler:

https://www.mathworks.com/help/releases/R2023b/simulink/ug/examine-solver-behavior-using-solver-profiler.html

[Alexander]

Hello @Yifeng Tang

Yes, you are right.

Thank you for the links. I'm studying it soon.