Convergence problem and the Mach problem in supersonic model

ranxiaoran · July 28, 2021, 06:02

Good afternoon, everyone.
I try to do the simulation about a two demensional supersonic chamber. The model is simple, it is like the picture below. There are two inlets, the larger one is called mainstream, the smaller one is called Cooled stream. The material is air. They have the same static pressure about 6200pa. The Mach of the mainstream is 2.4 and the Mach of the Cooled stream is 1.2. The total temperature is 291.44K(mainstream) and 233.15K(Cooled stream).
I use the pressure inlet for the mainstream and the mass flow inlet for the Cooled stream. I use isentropic formulation to calculate the total pressure.
Now I try to make this model convergence. However, I have meet difficulty on it. If I use the hybrid initialization, the convergence is good but the Mach of the inlet is wrong. If I use the standard one, the result cannot converge. If I use fmg initialization, I meet the same problem as hybrid initialization. I cannot solve this problem now.

The model is like this:stepped flow

LuckyTran · July 28, 2021, 09:51

When you use standard initialization, you have to specify the initial guess. You don't just click standard and let it run. Well you can, but the default guess will most likely be wrong and it will blow up in your face.

Regardless of what initialization method you use, you should get the same answer. If hybrid converged, then standard and fmg would have converged to the same solution unless... it is not converged. If you are unhappy with the result then something else is a problem, like your BC setting.

ranxiaoran · July 28, 2021, 12:27

Quote:

Originally Posted by LuckyTran

When you use standard initialization, you have to specify the initial guess. You don't just click standard and let it run. Well you can, but the default guess will most likely be wrong and it will blow up in your face.

Regardless of what initialization method you use, you should get the same answer. If hybrid converged, then standard and fmg would have converged to the same solution unless... it is not converged. If you are unhappy with the result then something else is a problem, like your BC setting.

Thank you for your reply. I am quite happy to get the advice from you. Maybe I need to change the initial guess🤔

July 28, 2021, 06:02	Convergence problem and the Mach problem in supersonic model	#1
ranxiaoran New Member iiran Join Date: Jul 2021 Posts: 21 Rep Power: 4	Good afternoon, everyone. I try to do the simulation about a two demensional supersonic chamber. The model is simple, it is like the picture below. There are two inlets, the larger one is called mainstream, the smaller one is called Cooled stream. The material is air. They have the same static pressure about 6200pa. The Mach of the mainstream is 2.4 and the Mach of the Cooled stream is 1.2. The total temperature is 291.44K(mainstream) and 233.15K(Cooled stream). I use the pressure inlet for the mainstream and the mass flow inlet for the Cooled stream. I use isentropic formulation to calculate the total pressure. Now I try to make this model convergence. However, I have meet difficulty on it. If I use the hybrid initialization, the convergence is good but the Mach of the inlet is wrong. If I use the standard one, the result cannot converge. If I use fmg initialization, I meet the same problem as hybrid initialization. I cannot solve this problem now. The model is like this:stepped flow

July 28, 2021, 09:51		#2
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,677 Rep Power: 66	When you use standard initialization, you have to specify the initial guess. You don't just click standard and let it run. Well you can, but the default guess will most likely be wrong and it will blow up in your face. Regardless of what initialization method you use, you should get the same answer. If hybrid converged, then standard and fmg would have converged to the same solution unless... it is not converged. If you are unhappy with the result then something else is a problem, like your BC setting.