For solving the unsteady Navier Stokes equation in a microfluidic channel on a rotating platform, i.e., a lab-on-a-CD platform, I use a time-dependent solver,BDF, for a three dimensional laminar flow of non-Newtonian Carreau blood. While my time stepping and mesh size work appropriately well for rotations at lower frequencies, I usually face an error at higher frequencies of rotation.
Repeated error test failures. May have reached a singularity.
Time : 0
Last time step is not converged.
Does anyone have a solution for such kind of an error?
I tried using the Generalized-alpha solver instead, which converges but takes a lot more time and is stated to be not that accurate.
I looked at your model. I changed the mesh so that in x-direction I reduced the number of elements to 50, increased in y-direction to 20 and z-direction to 16, with an arithmetic distribution so that at the boundaries elements were thinner.
First I solved the Study 1, steady-state. It took 3.5 mins with my 8GB RAM Windows 8 laptop. Then I used this solution as the inital value of the time dependent solution, and returned the relative tolerance back to the default value 0.01. Now the simulation took almost 9 mins.
Attached my file. I use Comsol 4.3b.
Hello Sir,
Thank you very much for taking the time to try solving my problem. I tried downloading the file and opening it in COMSOL Multiphysics 4.3a. But it says that the file is either corrupt or not valid. Is it because of the difference in the version of COMSOL used? Else, could you please take a snapshot of the places where you made these changes and upload it here. Would be a great help.
Thanks,
Monika
Hello Sir,
Thank you very much for taking the time to try solving my problem. I tried downloading the file and opening it in COMSOL Multiphysics 4.3a. But it says that the file is either corrupt or not valid. Is it because of the difference in the version of COMSOL used? Else, could you please take a snapshot of the places where you made these changes and upload it here. Would be a great help.
Thanks,
Monika
Yes, you cannot open model done with 4.3b with software 4.3a, but the changes are very small:
1. Mesh1 --> Mapped 1 --> Distribution 1: Predefined distribution type, Number of elements 10, Element ratio 10, Arithmetic distribution, Symmetric distribution.
2. Mesh1 --> Swept 1 --> Distribution 1: Fixed number of elements 50.
I attach a two snapshots, I wish you can read them.
I looked at your model. No data set is defined for the first plot, therefore nothing is displayed. But all your concentrations were zero!
Lasse
I used the comsol 4.3 .The same error appeared in my model.I try to change the mesh size,but it dosen't work .
how can i solve this problem
Thanks!
I do not understand your model, but you have problems with the units. Check your variables and functions.
Hi everyone,
I was wondering If someone could help me figure out the problem with my model. I am getting the same error as moika:
' Repeated error test failures. May have reached a singularity.
Time : 0
Last time step is not converged."
Thanks
@Lasse - I am happy to see one of the answer on changing the Mesh to Mapped. Its really interesting. Could you please explain the background / concept underlying on this?
Aftem meashing i found Repeated error test failures. May have reached a singularity.
Time : 0.03759313922805087
Last time step is not converged. i can not analyise. Any one can help me for solving this Comsol problem.
Hi everyone,
I also have a similar singularity problem for 'study 1' when I change the time interval from "range(0,10,2000)" to anything above 3400 seconds. I have adjusted the global parameters and tried to adjust the 'mesh' but it didn't work. The model is attached. Thanks in advance for your help.
Ekin
Hi Ekin,
I am also facing similar problem singularity problem at 1524 second when run study1 range(0,10,8000). Did you get any solution regarding this error
hi everyone
I have the same problem when using v4.3a to model dbd plasma actuator..can any one help me?
Hi,
A possible solution might be found here. https://www.comsol.com/support/knowledgebase/1127/
Try this and hope it woks.
Hi,
Use smoothing function if you have a step rise/fall in your boundary condition. This will avoid the singularity error.
Regards
S.Manikandan
Try increasing the mesh size at the singular point.
Generally, I would start the solution with a coarse mesh. Then gradually reduce the mesh size near the singular point.
Hope this helps.
- Badges
- Science topic