[siddu46]

hi every one,

i m doing CFD simulation using cfx solver, i m not able to get proper convergence for my solution.

my iteration run is showing 0.8 domain imbalance (P.mom imbalance (%) & U.mom imbalance (%)) of 0.8.

i am able to visualize residuals (P.mass.residuals) (ISO surface) by using output controls>results>output equation residual(enable)>all (variables) in cfx pre.

but, i am unable to visualize domain imbalance for the simulation (P.mom imbalance (%) & U.mom imbalance (%)).

kindly help me i am doing this exercise for the past 3 month.

[Gert-Jan]

Did you read: https://www.cfd-online.com/Wiki/Ansy...ated_questions
There you can find a lot of info to obtain better convregence.

But you mention your imbalance is high. Why? 0.8 means 0.8%. That is not so bad.
And do you know what the imbalance means? If so, then how do you want to visualise this? What do you expect to see?

Because an mass imbalance means that less mass is leaving your domain than entering. You might be able to visualise the velocity on inlet and outlet and have the luck to see a small difference (provided they have the same size and physical properties....)

[siddu46]

Ty GN for immediate reply,
1. I am following the document: https://www.cfd online.com/Wiki/Ansys_FAQ#Solver_related_questions

2. As for my understanding: Imbalance is the net effect of imbalance for a domain (flux: in and out) i.e., found at the boundaries (error difference between boundaries) , i.e., global imbalance.
residual is local imbalance i.e., for a controlled volume.( error due to the correction term in the numerical solution of the governing equation for a finite volume inside the domain).
so, it's only residual variables we can visualize through CFD post.
kindly, correct me if I am wrong.

difficulties of my simulation:
3. I am doing grid independence for my geometry, I am able to generate a solution for 0.5 Million Node (convergence up to 0.06 P.Mass.Imbalance) and 1 million node (convergence up to 0.06 P.Mass.Imbalance), with in 3000 iterations (physical timescale: 1e-04 up to 1000 iterations and 1e-03 further). but, for 2 million grid point I am unable to generate the solution (with the same time step). note: there is no mach number error in my solution.

kindly help me.

[Gert-Jan]

[QUOTE=siddu46;698026]Ty GN for immediate reply,
1. I am following the document: https://www.cfd online.com/Wiki/Ansys_FAQ#Solver_related_questions

Answer: Keep up the good work

2. As for my understanding: Imbalance is the net effect of imbalance for a domain (flux: in and out) i.e., found at the boundaries (error difference between boundaries) , i.e., global imbalance.
residual is local imbalance i.e., for a controlled volume.( error due to the correction term in the numerical solution of the governing equation for a finite volume inside the domain).
so, it's only residual variables we can visualize through CFD post.
kindly, correct me if I am wrong.

Answer: You are correct.

difficulties of my simulation:
3. I am doing grid independence for my geometry, I am able to generate a solution for 0.5 Million Node (convergence up to 0.06 P.Mass.Imbalance) and 1 million node (convergence up to 0.06 P.Mass.Imbalance), with in 3000 iterations (physical timescale: 1e-04 up to 1000 iterations and 1e-03 further). but, for 2 million grid point I am unable to generate the solution (with the same time step). note: there is no mach number error in my solution.

Answer:

In general, with a smaller mesh size, you will find more details in the flow, mostly leading to more instability. Then convergence is harder to get. Not much you can do.
I always use auto timescale. And I play around with the Timescale factor. Sometimes it helps to increase the factor, sometimes it is the other way around.

[siddu46]

hi GN,

thank you for your time:

1.I am getting a convergence of upto 0.013 % for 2 million nodes after 10000 iteration. i want to ensure my solution is correct, for that, i am checking res file for every 100 iteration and i am evaluating mass imbalance manually. i.e., M_in-M_out/M_in.
is this the correct methodology to ensure the solution.
2. I want to know, how cfx is evaluating p.mass.imbalance &u-mom.imbalance.

I followed cfx pre: it is given that, % imbalance = equation imbalance / maximum flow, how to evaluate equation imbalance for a particular domain. what is maximum flow in this. I want to understand how the following imbalance evaluation methodology in cfx:

U-Mom-san |
+--------------------------------------------------------------------+
Boundary : san_sid_int Side 1 6.1596E-05
Boundary : san_in -3.7074E-05
Boundary : san_w1 -2.4194E+02
Boundary : san_w2 -6.5975E-01
Boundary : san_w3 2.0234E+01
Domain Interface : san_sid_int (Side 1) 2.3665E+02
Domain Interface : per_san (Side 1) 1.5382E+03
Domain Interface : per_san (Side 2) -1.5525E+03
-----------
Domain Imbalance : -1.3714E-04

+--------------------------------------------------------------------+
| P-Mass-san |
+--------------------------------------------------------------------+
Boundary : san_in 6.6321E-01
Domain Interface : san_sid_int (Side 1) -6.6321E-01
Domain Interface : per_san (sid 1) -7.7000E-01
Domain Interface : per_san (Sid 2) 7.7000E-01
-----------
Domain Imbalance : 1.9730E-06

I think it is straight forward but even then I am not getting,

kindly help me to correct myself.