[Obnates]

I'm fairly new to FLUENT and turbulence modeling, so I need some help on a problem.

I'm trying to model a turbulent jet with Re ~ 2*10^4.

I am having problems getting my solution to properly converge, i.e. the centerline velocity is not linearly decaying like I might expect.

My computational domain is as follows:
North side: pressure outlet, 0 gauge pressure
East side: pressure outlet, 0 gauge pressure
South side: Axis
West side: top region is a pressure outlet, 0 gauge pressure; bottom region is a velocity inlet with V = 10 m/s

There is a jet wall separating the top and bottom regions (viewed from the west).

I'm currently using the realizeable k-e with enhanced wall treatment; my grid is more fine near the wall and coarse out in free space.

I may be approaching this entirely wrong -- what would proper boundary conditions be, and which turbulence model would work best?

Thanks.

[Ralf Schmidt]

Hi!

your BC do look right....

Check your yplus value on the wall... For enhanced wall treatment, I think, it should be between 1 and 5.

How do you determine the convergence of your solution? Do you take the residuals with standard convergence check?
Turn off that criteria - the standard values are set not low enough. And lowering the values by yourself might result in to low values (that cannot be reached....

Use an alternative criteria: a surface monitor of the (area averaged) axial (or radial) velocity on your axis (or on a point in your flow field). If that value does not change any more, your solution is (more or less) convergent!

Another point: have you set the problem to 2D axis symmetrical?

What is the distance between the jet and the boundaries? It should be something like 10 jet diameter perpendicular to the jet and more than 20 jet diameter in flow direction of the jet.

Best wishes
Ralf

[Obnates]

The y+ value on the wall is indeed <5.

I check convergence by a) monitoring the residuals with standard convergence check, b) monitoring the mass flow rate out of the free surfaces and c) monitoring the average velocity along the jet centerline.

I have set the problem to 2D axisymmetrical.

The distance between the velocity inlet and the jet face is 20 jet diameters.
The distance between the jet face and the east boundary is 30 jet diameters.
The distance between the jet wall and the north boundary is 5 jet diameters.

[Ralf Schmidt]

mhh.. ok.. that looks very well...

I do not really get your geometry... can you upload a picture of it?(including the grid)

And another idea: what are your reference values?

[Obnates]

A picture (not to scale) of my domain is shown.

I don't have a picture of the grid currently handy -- I could try to get one sometime later.

By "reference values" do you mean operating conditions?

[Ralf Schmidt]

so... that is what I thought about...
you distance between the jet wall and the north boundary might be a little low...

The reference values are given in -> report-> reference values.
It is a good idea, to compute them from the velocity inlet. They are used to calculate volume flow or heat transfer...

Have you scaled your grid?

can you give a plot of the axial velocity on the axis? What is the problem with these values?

Do you have a temperature change in your domain? What kind of material properties are you using?

[Obnates]

I considering making the distance between the jet and the north boundary 10 diameters, but someone I know suggested starting at 5 diameters and that he had success with that in the past.

What reference value in particular would be of interest in this case?

I have not scaled my grid -- how would I scale it, and what advantage does that have?

A plot of the axial velocity is attached -- from -0.3048, you can see the velocity increase as the flow goes down the pipe. Then, shortly after 0 (the jet exit), you can see the velocity decreases slightly (the potential core). Then it decays rapidly, then decays slowly. I expect to see (from self-similarity) the end of the potential core, and then a slow, constant (linear) decay.

There is no temperature change or heat transfer. The pipe is aluminum and the fluid is air into air.

[Ralf Schmidt]

Hi!

ok.. the reference values are not really interesting her... I grant... it was just an idea... The operating pressure should be atmosphere pressure unless your jet is in a vacuum...

The grid scaling is very important!! You create your grid in gambit WITHOUT any units. You just have dimensionless distances.

Fluent does import the grid WITHOUT any units as well. So, in Fluent you have to assign units to the grid (-> grid-> scale). The standard in Fluent is meters. So if you have created your model in mm, there will be a factor of 1000 to your geometry.

At the moment, I have no further idea, what is going wrong with your simulation....

[Obnates]

Re: grid scaling -- I created the geometry in meters, so would I still need to scale the grid?

I'm going to try to create a new grid with 10 diameters between the jet wall and the north boundary to see if maybe that makes a difference.

Thanks for all your help.