Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
Mass Fluxes Does not Match on Boundary, and Different Results from Different Solvers
Posted 2014年10月26日 GMT-4 21:35 Fluid & Heat, Chemical Reaction Engineering, Studies & Solvers 4 Replies
Please login with a confirmed email address before reporting spam
I'm currently modeling a mass diffusion problem, using Transport of Dilute Species and Laminar Flow modules. On the boundaries of two adjacent diffusion domains, I use the Stiff-Spring method as introduced in the dialysis example.
flux=M*(c2-k*c1), where k=1.
The first solver I chose was segregated solver. I've tried both iterative and direct methods. They converge to the same numerical value. However, the nominal total mass flux on the two sides of the boundaries are not the same. It means the mass balance is break.
In the attached excel, I copied the probe results from the segregated solver. Column D and E, Column F and G should have the same absolute value. But they differ a lot.
Then I tried fully-coupled solver. This time, the fluxes on the boundaries satisfy the mass balance. But the results it converges to is very different with that of segregated solver.
So is segregated solver not suitable in this problem? How to determine which solver to use?
Thank you!
Xiao
(Due to size limit of attachments, I deleted the results in the model attached.)
Attachments:
Please login with a confirmed email address before reporting spam
if I only solve the flow field for your model, I notice that there is influx at the outlet side of your model. This will cause problems when you also solve the concentration field, as the influx at that boundary is unspecified.
I would advise to make the outlet side longer, so that this influx is no longer present.
I would also include boundary layers in the mesh, and use laminar inflow conditions for the fluid.
Good luck with your model.
Frank
Attachments:
Please login with a confirmed email address before reporting spam
Thank you for your reply.
I improved my model according to your suggestions. The influx at the outflow is now eliminated.
However, the mass flux problem still exists using segregated group solvers. Mass fluxes on the left and right sides of a boundary do not match. Could it be the reason that M (assumed velocity) in the boundary condition not large enough?
flux=M*(c2-k*c1).
Attached model is basically the original model with a cut down size geometry. So it's easier to be solved than the original one. Using segregated solvers, the normal total fluxed on one boundary are 6.46E-10 and 1.28E-11. On the other boundary are 4.27E-18 and 4.24E-17.
Thanks,
Xiao
Attachments:
Please login with a confirmed email address before reporting spam
It seems that your inlet conditions for the air domains are off. So the selections are not properly assigned, and the direction of the flow is in the positive z direction instead of downwards.
Furthermore it can be a good idea to make the air inlet and outlet slightly longer, so that the conditions on the corners do not conflict.
On the backflow in the fluid part, here you need to make the model larger on the x axis, not in the z direction. Alternatively you can enable stokes flow to simplify the flow, but with a reynolds number of about 700 this is not really valid.
Note that there will always be slightly different results depending on what solver strategy you choose. But after mesh convergence and reducing the value of your absolute and relative tolerances, you should obtain accurate and similar values.
Best regards,
Frank
Attachments:
Please login with a confirmed email address before reporting spam
Xiao