CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   Convergence Problem in Axisymmetric Periodic Flow (https://www.cfd-online.com/Forums/fluent/129597-convergence-problem-axisymmetric-periodic-flow.html)

atheresia February 8, 2014 11:12

Convergence Problem in Axisymmetric Periodic Flow
 
Hi everyone,


I am modelling a periodic flow for a 3D pipe, using an axisymmetric model.
The flow is to be in the Z direction, with half-width (or radius) of 3 m.

The Reynolds number is 34,132 and the working fluid is air.
This gives my inlet velocity of 0.081 m/s.
I set the turbulence intensity as 4.3% and turbulent length scale of 0.42.

I have a problem, however, in obtaining convergence.
My current convergence criterion is 1e-06 for all the variables.

With the initialization values below:
http://i59.tinypic.com/34hxg82.jpg

My convergence plot is:
http://i61.tinypic.com/2qxnrxi.png

I have tried to run a full 3D pipe periodic flow as well. For my initial mesh, I managed to obtain convergence, however, as I refine it further (by increasing the bias factor from 10 to 20), now it won't converge.

Not sure what's going on! So any comments/feedbacks will be greatly appreciated!

Thanks so much in advance for your help!

flotus1 February 9, 2014 04:13

2D axisymmetric domains in fluent must have the x-axis as symmetry axis.
Additionally, for translational periodic flows with a very short streamwise extent, it is not unusual to take 10e5 or even more iterations to converge.
10e-6 for the residual values may still be too loose as a convergence criterion in this case. Make sure to double-check convergence with some other quantity like wall shear stress.

atheresia February 9, 2014 17:03

Quote:

Originally Posted by flotus1 (Post 474103)
2D axisymmetric domains in fluent must have the x-axis as symmetry axis.
Additionally, for translational periodic flows with a very short streamwise extent, it is not unusual to take 10e5 or even more iterations to converge.
10e-6 for the residual values may still be too loose as a convergence criterion in this case. Make sure to double-check convergence with some other quantity like wall shear stress.

Thanks so much for your reply Alex!

I've just tried it in xy plane and it converges now!! Thank you!

I am not sure, however, how to 'double-check convergence' with other quantity like wall shear stress.. I'm still new to CFD so I would really appreciate it if you would share some guidelines/resources on this? :-)

Many thanks!!

flotus1 February 10, 2014 03:00

The basic idea is that you should never judge convergence based on the value of some residuals alone.
They are a good indicator, but sooner or later you will run into trouble if you rely on them too much.
The absolute minimum for a simulation like yours is to set up a monitor point to monitor the average wall shear stress at the pipe wall.

These two links have lots of valuable information on this topic:
http://www.cfd-online.com/Forums/flu...nvergence.html
http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F


All times are GMT -4. The time now is 01:18.