CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   What is Floating pt error?? (

JJ April 14, 2001 06:09

What is Floating pt error??
Hi there,

I was prompted with the following error message by Fluent when I try to plot concentration contour on one particular plane in the domain,

Error message: "Floating point error : overflow"

What does it mean? I am able to plot and do other things except ploting on that plane! The solution is already converged.



Scott W April 16, 2001 15:30

Re: What is Floating pt error??
I've seen that occur when a plane lies just outside the domain. For example if you have a boundary at x=0, and then create a surface of constant x at x=0. This should lie right along your domain boundary. However, due to rounding errors, Fluent thinks the domain is at x=0+ (such as x=0.00000000000001) and you are trying to display outside that domain at x=0.00000000000000.

JJ April 17, 2001 01:42

Re: Can the solution be trusted?
Hi Scott,

Thanx for the answer.

Since this seems to be a prob with the grid that I have drawn, does it imply that the solution obtain with it could not be trusted?

Cheers JJ

Scott W April 17, 2001 09:39

Re: Can the solution be trusted?
There are two ways to trust your solution: 1) Compare parts of it to experimental data. If they match within your tolerance, then you can trust your solution. 2) While not quite as good, compare your answer to simple analytical solutions. If there are major discrepancies then either your solution is wrong, the analytical solution doesn't model reality, or both.

Unless you have the simpliest of all geometries, the mesh will not exactly mesh the geometry you created. For example, draw a curve on a piece of paper and hand-mesh it with rectanges. You have either 1) mesh points outside the curve domain or 2) the mesh doesn't fully reach the edge of the domain. The finer the mesh, the better the approximation, but you can never fully mesh the curve until Fluent lets us have curved edges on our cells. It is up to you to compare your CFD results to something else to determine if this is a problem.

JJ April 18, 2001 03:43

Re: thanx!!!
Scott, thanks a lot for your answer, they were a great help!

Cheers JJ

All times are GMT -4. The time now is 18:52.