CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   Error: Floating Point Exception. How do I fix this? (https://www.cfd-online.com/Forums/fluent/228938-error-floating-point-exception-how-do-i-fix.html)

Ryan T July 20, 2020 23:10

Error: Floating Point Exception. How do I fix this?
 
3 Attachment(s)
Quote:

Divergence detected in AMG solver: k Stabilizing omega to enhance linear solver robustness.
Stabilizing omega using GMRES to enhance linear solver robustness.

Divergence detected in AMG solver: omega
Divergence detected in AMG solver: k
Divergence detected in AMG solver: omega
Divergence detected in AMG solver: k
Divergence detected in AMG solver: omega
Error at host: floating point exception

Error at Node 0: floating point exception
Error at Node 1: floating point exception
Error at Node 2: floating point exception

===============Message from the Cortex Process================================

Compute processes interrupted. Processing can be resumed.

================================================== ============================

Error: floating point exception
Error Object: #f

Calculation complete.
My Fluent settings are

Density Based
Energy (On)
Standard k-omega model
Air is set to Ideal Gas
Standard Initialization from the Inlet.

I attached a picture of the mesh along with the mesh details section and inflation settings. Its supposed to be a supersonic nozzle leading into a wind tunnel test section to simulate a supersonic wind tunnel test section. I have 2 edge sizings as well with an element sizing of 0.001 m. I have the solver set to Double Precision. I've also tried Hybrid Initialization along with other turbulence models and its given me the same errors.

How do I fix this? This might be an issue with the mesh because meshing isn't my strong suit.

I'm on version 2019 R3

Guvennergiz July 21, 2020 02:36

1 Attachment(s)
Hello Ryan,
1st of all can you check your mesh quality with the figure below (for min ortho. quality > 0.1 or max. skewness <0.95) ?
Best regards,
Güven

Ryan T July 21, 2020 05:22

1 Attachment(s)
Quote:

Originally Posted by Guvennergiz (Post 778308)
Hello Ryan,
1st of all can you check your mesh quality with the figure below (for min ortho. quality > 0.1 or max. skewness <0.95) ?
Best regards,
Güven

Hello, in Fluent the report quality says

Quote:

Mesh Quality:

Minimum Orthogonal Quality = 1.29257e-01 cell 9333 on zone 2 (ID: 81407 on partition: 0) at location ( 4.44510e-01 -4.14017e-02)
(To improve Orthogonal quality , use "Inverse Orthogonal Quality" in Fluent Meshing,
where Inverse Orthogonal Quality = 1 - Orthogonal Quality)

Maximum Aspect Ratio = 7.02383e+02 cell 24136 on zone 2 (ID: 1384 on partition: 1) at location ( 7.32540e-02 4.27523e-02)
I've also attached what the mesh metrics says. I'm not quite sure how to evaluate it.

EDIT: Yeah report quality isn't giving me the mesh skewness. I'm on version 2019 R3.

Guvennergiz July 21, 2020 05:40

1 Attachment(s)
Hello Ryan,
0.12 ortho. quality is in the "Bad" section on sprectrum. You can try to improve your ortho. quality.
Did you setup "Fluent with meshing"? If it is, unfortunately i do not have enough information on it to help you. But if you set the mesh section differently, please open your mesh and view your skewness and orthogonal quality. When you select these sections, a graphic will appear below showing the distribution of the elements. When you click on the columns that showing low quality elements, it will show you their place on the existing mesh. After that, you need to use various method-sizing-inflation settings to improve the quality of those elements.
There are several tutorials on youtube about that topic.
Hope that it helps to you.
Best regards,
Güven

Ryan T July 21, 2020 05:46

2 Attachment(s)
Quote:

Originally Posted by Guvennergiz (Post 778332)
Hello Ryan,
0.12 ortho. quality is in the "Bad" section on sprectrum. You can try to improve your ortho. quality.
Did you setup "Fluent with meshing"? If it is, unfortunately i do not have enough information on it to help you. But if you set the mesh section differently, please open your mesh and view your skewness and orthogonal quality. When you select these sections, a graphic will appear below showing the distribution of the elements. When you click on the columns that showing low quality elements, it will show you their place on the existing mesh. After that, you need to use various method-sizing-inflation settings to improve the quality of those elements.
There are several tutorials on youtube about that topic.
Hope that it helps to you.
Best regards,
Güven

Yeah I attached the pics of what mesh and skewness are in the Mesh itself. I'm not exactly sure what you mean by clicking the columns to show low quality elements from the graphic.

Guvennergiz July 21, 2020 05:50

Hello Ryan,
There is a huge difference between "ortho. quality in Fluent" and "ortho. quality in meshing" as you can see. Did you make polyhedra mesh in Fluent?
Best regards,
Güven

Ryan T July 21, 2020 05:51

Quote:

Originally Posted by Guvennergiz (Post 778334)
Hello Ryan,
There is a huge difference between "ortho. quality in Fluent" and "ortho. quality in meshing" as you can see. Did you make polyhedra mesh in Fluent?
Best regards,
Güven

No I made the mesh in meshing. But what is the difference between orthogonal quality in meshing and fluent? Why are they giving different values?

Guvennergiz July 21, 2020 06:06

Hello Ryan,
That difference usually show up when you make polyhedra mesh in fluent (e.g you've created your mesh then opened Fluent, you can convert your mesh to polyhedra with a single button in Fluent and it can not be converted back).
Can you try to improve your mesh quality just a bit? After updating mesh, please check quality in Fluent again.
Best regards,
Güven

Ryan T July 21, 2020 06:12

2 Attachment(s)
Quote:

Originally Posted by Guvennergiz (Post 778334)
Hello Ryan,
There is a huge difference between "ortho. quality in Fluent" and "ortho. quality in meshing" as you can see. Did you make polyhedra mesh in Fluent?
Best regards,
Güven

So I changed my method from "All Triangles" to "MultiZone Quad/Tri Method" to yield the skewness and orthogonal quality values I attached. The values are improved over All Triangles Method.

The Report Quality in Fluent says
Quote:

Mesh Quality:

Minimum Orthogonal Quality = 3.82862e-01 cell 28028 on zone 2 (ID: 70201 on partition: 2) at location (-1.66396e-01 -2.58179e-01)
(To improve Orthogonal quality , use "Inverse Orthogonal Quality" in Fluent Meshing,
where Inverse Orthogonal Quality = 1 - Orthogonal Quality)

Maximum Aspect Ratio = 1.37639e+03 cell 0 on zone 2 (ID: 1 on partition: 2) at location (-1.66396e-01 2.58179e-01)
But it still gives me the same error.
Quote:

Divergence detected in AMG solver: k Stabilizing omega to enhance linear solver robustness.
Stabilizing omega using GMRES to enhance linear solver robustness.

Divergence detected in AMG solver: omega
Divergence detected in AMG solver: k
Divergence detected in AMG solver: omega
Divergence detected in AMG solver: k
Divergence detected in AMG solver: omega
Error at host: floating point exception
Error at Node 0: floating point exception
Error at Node 1: floating point exception
Error at Node 2: floating point exception

===============Message from the Cortex Process================================

Compute processes interrupted. Processing can be resumed.

================================================== ============================

Error: floating point exception
Error Object: #f

Guvennergiz July 21, 2020 07:08

Hi Ryan,
So we can continue with 2nd part: time step (ofc if you certainly sure your IC and BC).
I do not know what your time step size right now. Can you decrease your time step size and try again?
Btw, you are getting error while calculating right? Not initializing?
Best regards,
Güven

Ryan T July 21, 2020 07:15

Quote:

Originally Posted by Guvennergiz (Post 778343)
Hi Ryan,
So we can continue with 2nd part: time step (ofc if you certainly sure your IC and BC).
I do not know what your time step size right now. Can you decrease your time step size and try again?
Btw, you are getting error while calculating right? Not initializing?
Best regards,
Güven

I have my time set to Steady instead of Transient right now, but I'll try Transient.

EDIT: I have tried a time step size of 0.5 sec and 1 sec and it is still giving me the same error, and yes the errors are during the calculation process. Not initialization.

KruX July 21, 2020 07:36

Your max. Aspect Ratio is very high. Maybe a structured mesh would improve your overall mesh quality.

Ryan T July 21, 2020 15:42

Quote:

Originally Posted by KruX (Post 778346)
Your max. Aspect Ratio is very high. Maybe a structured mesh would improve your overall mesh quality.

Maybe but I'm trying to keep the inflation to simulate the boundary layer.

KruX July 21, 2020 15:52

You can do the same with a bias in the sizing option for a structured mesh. I think the overall AR should be below 5. For the cells in the boundary layer 30 is okay, but you have a maximum AR of 700 which is way too high.

AlexanderZ July 22, 2020 03:13

sharp angles near the boundary is a problem definitely.

is your model axis symmetric? in that case you need just half of model
switch to polyhedral mesh to increase quality and decrease number of elements

Ryan T July 22, 2020 17:08

Quote:

Originally Posted by KruX (Post 778403)
You can do the same with a bias in the sizing option for a structured mesh. I think the overall AR should be below 5. For the cells in the boundary layer 30 is okay, but you have a maximum AR of 700 which is way too high.

Yeah I'll look into that then. I wasn't aware my AR was too high.

Quote:

Originally Posted by AlexanderZ (Post 778428)
sharp angles near the boundary is a problem definitely.

is your model axis symmetric? in that case you need just half of model
switch to polyhedral mesh to increase quality and decrease number of elements

No my model isn't axis-symmetric. What's the option for a polyhedral mesh? In meshing there seems to be only quadrilaterals, triangles, or multizone quad/tri. And yeah right at the start of my mesh when its sharp seems to also be the points of highest skewness and lowest orthogonal quality.

Ryan T July 30, 2020 19:52

Quote:

Originally Posted by KruX (Post 778403)
You can do the same with a bias in the sizing option for a structured mesh. I think the overall AR should be below 5. For the cells in the boundary layer 30 is okay, but you have a maximum AR of 700 which is way too high.

My new structured mesh has a max AR of 8.8634 and an average AR of 2.6413. Do you mean that the average AR should be less than 5 or the max AR should be less than 5?


All times are GMT -4. The time now is 17:52.