Hello,

I'm trying to model a plain underexpanded jet (still), and I can't get the iterations started. Fluent keeps coming back with floating point exception, (underflow). I don't know what it's complaining about!.

I've tried reducing the skewness, checked the grid, re-entered the boundary conditions (trying to start off with a jet at a very low NPR, without turbulence), no joy.. am I grasping at straws or is it something blindingly obvious?

please help

Thanks

Your model is diverging.

The underexpanded jet is almost impossible to converge using Fluent.

Make sure you are using the coupled solver, and maybe some kind of initial guess.

Can you explain "Underexpanded Jet"? Thanks!

Underexpanded jet = a compressible gas flow where the pressure in the jet is substantially higher than the pressure of the region into which it is discharging - leads to lots of shock and expansion waves etc

(1). That's right. (2). Or, (p,ambient/ p,jet exit) < 1. (3). (p,jet exit) is the nozzle exit pressure but still inside the nozzle. (4). Further expansion will occur in the jet, downstream of the nozzle exit, until the pressure at the boundary of the jet matches the ambient pressure. In between, shocks and expansion waves will form in the jet (or plume).

Thanks,

Managed to get the model started, stripped it down completely to a laminar, incompressible flow at a very low NPR to begin with and gradually increased the NPR as the solution progressed.

Have you tried to patch a non-zero initial guess for the velocity field ? Please contact your support engineer if the problem doesn't go away.

Thanks for the reply

Non-zero intial guesses also prompted the error, It wasn't until I took stripped down the model that it was able to start. Unfortunately I don't think I have a support engineer, I'm a post-grad and the University has an academic license to which I am not the primary licence holder, queries to fluent (europe) support are very slow in replying (if at all) because an academic license is has low priority. Hence I look to the fluent forum for possible help.

[sheikh nasir]

Hello.
I am getting floating point error:invalid number in my thesis. Ie train moving in tunnel. Can any help me. My email is sheikhnasir39@gmail.com
thanks

[chngyian]

One reason could be that the factors are over-relaxed. So if you are confident that BCs, model and solver are appropriate, try reducing the relaxation factor in solution control.

[Omish]

Originally Posted by Ricky Wong
;93808

Thanks,

Managed to get the model started, stripped it down completely to a laminar, incompressible flow at a very low NPR to begin with and gradually increased the NPR as the solution progressed.

What does NOR stand for?

[Omish]

Originally Posted by Ricky Wong
;93808

Thanks,

Managed to get the model started, stripped it down completely to a laminar, incompressible flow at a very low NPR to begin with and gradually increased the NPR as the solution progressed.

What does NPR stand for?

[niazgdc]

Originally Posted by Omish

What does NPR stand for?

i have the following error in my simulation, kindly if u hav any remedy,post it
# Divergence detected in AMG solver: pressure correction -> Turning off correction scaling!
# Divergence detected in AMG solver: pressure correction -> Increasing relaxation sweeps!
# You may try the enhanced divergence recovery with (rpsetvar 'amg/protective-enhanced? #t)

Divergence detected in AMG solver: pressure correction
Divergence detected in AMG solver: pressure correction
Divergence detected in AMG solver: pressure correction
Divergence detected in AMG solver: pressure correction
Divergence detected in AMG solver: pressure correction
Divergence detected in AMG solver: pressure correction
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

Error at Node 3: floating point exception

Error at Node 4: floating point exception

Error at Node 5: floating point exception

Error: floating point exception

Error Object: #f

[AM5840691]

hI

i AM DOING CFD SIMULATION FOR SINGLE CHANNEL PROTON EXCHANGE MEMBRANE FUEL CELL.

I AM RECEVING """""" FLOAT POINT EXCEPTION ERROR """" DURING CALCULATIONS..

cAN ANYBODY HELP REGARDING THIS?? HOW CAN I FIX THIS?

[AM5840691]

I m also receving the same error.......

[Rockfeaver]

you may change your AMG solver settings:

go to Solution Controls --> Advanced

You may add Pre-Sweeps in both scalar and coupled parameters.
Also I advise to use ILU smothing.

Apart from that, you may change Cycle type for Flow, TKE, SDR, Enregy, and Radiation model to either V- or W-Cycle. Also you could add BCGSTAB Stabilization Method.

I advise you to use earlier Fluent version, such as 14.5, and do set ups and calculations there.

Be sure that you use the same fluent version all the time!

P.S.: if this doesn't work, try to do hybrid initialization without patching, and run couple of iterations with low URFs.

Hope it helped.

[wkx410]

hello
I'm also doing simulations of underexpanded flows, so I've had a similar problem, and my solution is use Density-Based,formulation is implicit,Flux Type is AUSM,if caculate for short step,It will not be a problem.

[tatiana33]

Hello,

Has anybody come across the reason and the possible alternatives to solve this issue?
I am working with P-based, mixture model and a UDRGM. I am obtaining this error since time step 1.

Any help will be greatly appreciated!

[BasilVarghese]

I have been doing simulation on natural convection in heat sink in transient state.
I have given time step size as 5sec, no. Of time steps as 20,no.of iterations per time step as 50.
In monitors plot, I plotted static temperature on fin wall as convergence but even after 3000 iterations, temperature line is not becoming constant and residuals are also decreasing. No other divergence error occurs. But temperature graph is increasing steadily
How can i make it converge?

[vinerm]

The value of the time-step is critical in most of the transient simulations and need to be chosen properly. 5 s is a very very large value for time-step. Depending upon Gr number, it could be anything from around 0.1 s to all the way down to 1e-6 or even 1e-7 s. But 5 s is not possible. So, use such a time-step so that solution within every time-step converges in less than or equal to 30 iterations.