[FliegenderZirkus]

I have a two-fluid atomizatinon nozzle of the airblast/air-assist type as per this picture: https://imgur.com/xgoG975
The airflow is sonic (supply pressure is over 4 bar) and the ratio of air mass flow to water mass flow is about 100. Given the high air velocity, I guess the nozzle is "air-assist" rather than "airblast", but that's probably not too important.
I would like to use the built-in atomization model in Fluent but I'm not sure how to model the air intake. My thoughts so far:

1. Model flow as compressible and include the actual nozzle internal geometry. This should be most accurate, but is expensive and could prove difficult to achieve convergence. My main area of interest is the large chamber itself, not the immediate nozzle vicinity.
2. Model flow as compressible but only model the air inlet as a flat annulus
3. Model flow as incompressible and only model the air inlet as a flat annulus. The velocities will be wrong, but will this impact the DPM?

One of the inputs to the airbast DPM model is the relative velocity between the liquid and the gas, which is related to the thoughts above. I'm wondering how to calculate this?

The air is actually swirling a bit due to the shape of the passage inside the nozzle, but I'm leaving that complexity out for now. Thanks for any tips!

[vinerm]

Air flow has to be compressible if Ma is more than 0.3. And it does affect DPM.

1. If the nozzle geometry is simulated, then atomizer model is not required. You can use VOF-to-DPM transition model.

2. If you want to use inbuilt atomizer models, then the actual geometry of the nozzle is not important from the DPM view point. However, you need to include its effects on the boundary conditions for air inlet. So, if gas is swirling because of the nozzle, then you must provide that swirl to the air.

3. Incompressible air only if Ma < 0.3. Rest of the setup could be same as either 1 or 2.

Since the flow channels for air and liquid are different, you can calculate velocities based on mass flow rate and compressibility. Then you can calculate relative velocity. Users also use this as a tuning parameter to match data with the experiments.

[FliegenderZirkus]

Thanks for the quick reply.

If the nozzle geometry is simulated, then atomizer model is not required.

I understand this for the liquid phase. The options here are: either include the nozzle internal geometry and use VOF (or VOF-to-DPM), or use DPM utilizing the atomizer model. But on the air side, is it not an option to model the nozzle internals so as not to have to make assumptions about the air inlet BC? Clearly, the particle injection itself is not impacted by how the air intake is modelled, but as soon as the particles are released, they will be influenced by the surrounding air flow, which could be sensitive to the air inlet BC. I'm not saying I want to do this, just asking whether it's an option to consider. Ideally I'd have a simple circular or annular inlet at the wall and prescribe velocity vectors, but I'll need to come up with arguments to support such a simplification.

Regarding the air velocity calculation, I know how mass flow changes with pressure difference across the nozzle from the manufacturer's catalogue. It turns out this is a linear function - doubling the pressure doubles the flowrate. Am I right that this indicates choked flow? I tried using these formulas:
https://en.wikipedia.org/wiki/Orific...pressible_flow
and I managed to tune the discharge coefficient to match the catalogue data reasonably well. However, I'm still unsure how to use the fact that Ma=1 to get the air velocity. The speed of sound at nozle exit still depends on temperature, which I don't have, or do I?

Just a remark: VOF-to-DPM (let alone pure VOF) is too expensive for this simulation so that's something I already ruled out.

[vinerm]

If you are not interested in using VOF-to-DPM, then it does not matter how complex the geometry of the nozzle is. All that matter is the annular area where the air comes out. Its not a simplification, there are no assumptions. You just have to apply correct velocity vector. You don't need to know exact temperature, just an order of magnitude, which you can get from the application of the nozzle. At STP, air velocity higher than 100 m/s implies compressible flow. If it is lesser than that, then keep it incompressible.