CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Error #001100279 Floating point exception: Zero divide (http://www.cfd-online.com/Forums/cfx/117022-error-001100279-floating-point-exception-zero-divide.html)

ch_ohm April 30, 2013 13:55

Error #001100279 Floating point exception: Zero divide
 
Dear all,

Now I have simulated the warehouse size 20*20*30 meters. It is quite big. I need to simulate and monitor the air temp in the warehouse (10 h). I work with transient model with time step = 1 hour. But I face with error as above

Error #001100279
Floating point exception: Zero divide

The software is Ansys 14 CFX. CPU core i5 with RAM 6 gb

Please help .

Thank you

evcelica April 30, 2013 17:25

What influenced you to choose a 1 hour time step? That's probably several orders of magnitude too big!

ghorrocks April 30, 2013 19:36

FAQ: http://www.cfd-online.com/Wiki/Ansys...do_about_it.3F

But Erik is right - your time step is miles too big.

RicochetJ May 1, 2013 05:09

In your warehouse do you have buoyancy driven flow?

In the CFX manual it tells you how to estimate the time step.

If you're doing transient simulations you might want to consider adaptive time stepping. Set the maximum time step to something like 1hr and the minimum to something small like 1e-8s and try to converge within 5-10 loops.

ghorrocks May 1, 2013 07:01

Agreed, adaptive timestepping is the best choice - especially for beginners who have no idea of what timestep is really required. I would say that half the convergence problems we see on this forum are just from too large a timestep.

For most simulations 3-5 coeff loops is more appropriate. 5-10 is only required for models with tricky physics to converge suck as multiphase (but for most multphase models I still use 3-5).

RicochetJ May 1, 2013 07:49

Quote:

Originally Posted by ghorrocks (Post 424398)
Agreed, adaptive timestepping is the best choice - especially for beginners who have no idea of what timestep is really required. I would say that half the convergence problems we see on this forum are just from too large a timestep.

For most simulations 3-5 coeff loops is more appropriate. 5-10 is only required for models with tricky physics to converge suck as multiphase (but for most multphase models I still use 3-5).


Would you consider buoyancy driven flow to be "tricky".

ghorrocks May 1, 2013 08:05

No. 'Tricky' means in this case where complex coupling between phases/physics is required; such as 2-way coupled particle tracking, eularian particle models, chemistry, radiation modelling where the fluid flow interacts 2-ways with the radiation modelling and things like that.

Buoyancy is still single phase so that is best done with short time steps converged quickly - so 3-5 coeff loops per iteration.

ch_ohm May 4, 2013 05:47

Thank you all. I will try follow your suggest. Thank you again.


All times are GMT -4. The time now is 04:06.