[rw511]

Hello there,

This being my first post and all please treat me kindly. I'm trying to use CFX to simulate an experiment I saw, which I have since learned is a de Laval nozzle flow.

I have read a lot of the posts on nozzle flow, I have read the only supersonic flow tutorial CFX has (flow over a wing), have searched numerous sites but to no avail, so am hoping someone here can help.

The conditions given in the experiment specify:

1) the temperatures of both the air flowing into the nozzle and the quiescent air the jet is projecting into,
2) the pressure at the nozzle's exit as well as the stagnation pressure,
3) the flow speed achieved (Mach 3)
4) and numerous dimensions of the setup (which I had no problem implementing and meshing.

I managed to get a supersonic flow working on just a plain round jet flow but cannot for the life of me get even the simplest supersonic flow (oxy-moron right there) to work on my jet's geometry. I've tried fiddling with the timescale, starting from a low velocity and cannot get a solution for love nor money. Any input anyone has towards this (such as is this best done as multiphase to take two temperatures into account?) or link to a thread that I have foolishly overlooked, would be appreciated.

Rich.

[ghorrocks]

This being my first post and all please treat me kindly.

I promise to be gentle.

Try local time scale factor to help in convergence. Also is your mesh high quality? supersonic flows really need a good mesh. This FAQ covers the main issues http://www.cfd-online.com/Wiki/Ansys...gence_criteria - so if nothing else works run it transient. That just about always works, but takes a long time.

What do you mean by the multiphase approach you refer to? I do not know what you mean.

[rw511]

First off, thank you for the reply.

I have since had some success as you correctly suggested with a transient scheme and shall pursue this route, but have switched to using FLUENT.

The multi-phase approach of which I mentioned was modeling a chamber of initially still air at the temperature set out in the experiment, then releasing another phase of air into the nozzle at the temperature also specified. I have since come to realise I was perhaps over complicating things looking at it this way and have got a very nice looking shock when viewing my results as a transient scheme.

As regards a fine mesh, I started with what I considered to be a fine mesh and implemented a mesh adaption based on the absolute pressure gradients, which seemed to make the CFX solver crash. For this reason I switched over to FLUENT's 2D axisymmetric model with a different mesh and have done better.

Although my results seem reasonable, getting the continuity residual below 1e-3 is still not happening, but for the purposes with which I shall use the data, I am more than happy with the outcome.

Thank you again for taking the time to help.

Richard.

[ghorrocks]

a chamber of initially still air at the temperature set out in the experiment, then releasing another phase of air into the nozzle at the temperature also specified

This is not a multiphase simulation! This is a simple single phase simulation with gasses at different temperatures. Phase (as in "multiphase") is solid, liquid or gas. If you only have one phase (even if they are different gases) the simulation is not multiphase.

I started with what I considered to be a fine mesh

One of the most common beginner mistakes is to generate what they think is an "adequate" mesh then assume the mesh is not the cause of inaccuracies and problems down the track. More often then not a mesh generated by a beginner is not adequate. This FAQ is just the start in considering making your simulation accurate: http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F