hi, CFX experts

I am modeling some different nozzle designs and trying to predict the subsquent spray pattern (e.g., spray angle, plume shape). As a first step, my 2D computational domain includes a 1cm straight nozzle channel (120um in diameter) connected with a wide open air space. The liquid passed throught the channel and blasts into the open air as a jet. I tried several Eulerian-Eulerian type simulations, but none of them works.

I understand there is a spray model include spray breakup model for particle tracking by predefining the spray angle in CFX10. However, our simulation goal is to simulate spray angle. Does any person have idea to do such a kind of simulation? My email: stonebrook2003@hotmail.com

Thanks a lot!

The sort of problem needs to be modelled as an Eularian--Eularian multiphase flow. You will also need to include surface tension effects. The simulation will be computationally expensive due to the fine grid needed to capture the behaviour of multiple small droplets.

I would suggest looking for some validation papers before you attempt to do this from scratch.

hi, Joe, thanks a lot. I am working on literature review. Actually, I used to do a lot of Eulerian-Lagrangian spray simulations (blob model, spray breaup model etc) with CFX10 by predetermining the orifice conditions and got very good results. But this time, the simulation involves internal nozzle, so things become different.

I know, I may need to investigate more before asking the following questions, but for quick guidance purpose (I am sort of in rush), I appreciate your answers on them.

1. Should I model air as continous flow and liquid as dispersed fluid Vs. both air & liquid as continous flows? I incline to model liquid as disperse fluid, but I need to input a mean diameter which is unknown since I want to simulate the droplet diameter distribution.

2. Musig is an option I should try?

I also appreciate any suggestions on assumptions to simplfy the problem.

Thanks a lot!

I am not an expert on spray breakup but I would have thought that the only sensible choice is to model both fluids as continuous. Surface tension is critical to correctly predicting droplet transition and you can't apply that to a dispersed phase.

Musig is only applicable to fluids that are already in droplet form.

I have heard second-hand comments that CFXs surface tension model isn't particularly good, so I would certainly try to get a relevant paper before getting too deeply involved in trial and error.

You have to use very small timesteps because I think the effects (surface tension, contactangle etc.) you want to model are not implemented implicitly, but explicitly.

I would consider using a VOF method, which is not yet available in CFX. Try Fluent of STAR.

Astrid