CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Additional Nucleation for already wet steam (

Adam December 1, 2008 05:31

Additional Nucleation for already wet steam
I am trying to simulate the Laval nozzle of Moore et all (1973) where wet steam enters at the inlet and undergoes another nucleation in the nozzle.

Previous simulations involved only nucleation of originally dry steam at the inlet. They all ran fine, but the non-equlibrium steam nozzle always crashes when I simulate already wet steam.

I have tried 2 main ways

1 - Specifying wetness fraction and droplet diameter for one liquid phase at the inlet. There is not a great deal of sensitivity for droplet diameter.

2 - Running as if it is dry steam at the inlet, and simulating nucleation. Then restarting with 2 liquid phases, which now includes the wet phase from the inlet with an associated droplet diameter.

Both crash within say 20 interations with a message such as

+--------------------------------------------------------------------+ | ERROR #004100018 has occurred in subroutine FINMES. | | Message: | | Fatal overflow in linear solver. |

Any ideas on how to proceed?



Glenn Horrocks December 1, 2008 21:49

Re: Additional Nucleation for already wet steam

Are you using V12? There is some new stuff and a tutorial example in V12 for non-equilibrium steam flow in a turbine. If you have not seen it I strongly recommend you contact CFX support and get a copy of the beta of V12 and the tutorial manual.

Glenn Horrocks

Adam December 2, 2008 03:55

Re: Additional Nucleation for already wet steam
OK, I'll try to get a copy.

I need to use the MUSIG function.



Bart Prast December 15, 2008 05:26

Re: Additional Nucleation for already wet steam

as you might now, nucleation and droplet growth result in some very stiff source terms. Unless the linearization of these source terms is done realy well (I doubt it on CFX). You're first step is to go for really small time steps (1E-6), untill your nucleation peak is established. In our experience modelling nucleation with a implicit solver is a bit slow as really small timesteps are required (CFL<<1). An explicit solver in CFX would be nice.

I don't know whether you use it but avoid ASM (algebraic slip model). It uses turbulent diffusion for particle dispersion.


HekLeR January 19, 2009 00:15

Re: Additional Nucleation for already wet steam
The MUSIG and ASM features do not work with the nucleating gas model in CFX. To model polydispersed flow you can only add more phases with additional nucleation sources, and, unfortunately breakup & coalescense are not accounted for... allthough that is another R&D topic for this type of flow.

All times are GMT -4. The time now is 01:06.