[hsieh]

Hi,

We did one experiment by comparing a laminar jet and a turbulent jet. Problem setup is simple. An empty cup with the top open. A nozzle sitting on top of the cup opening at the cup axis. Liquid jet comes of the nozzle. The diameter of the cup is about 20 times of the nozzle. The length of the nozzle is about about 20 times of the nozzle inner diameter. In the laminar case, the inlet of the nozzle is connected to a long straight tube. For the turbulent case, the inlet of the nozzle is connected to a tubing with several 90 degree bends. For both cases, the flow rate at the inlet of the nozzle are the same.

I used interFoam for the laminar case. Perfect, numerical results matches the experiment results -> a very steady stream coming out of the nozzle. No liquid breaks up before hitting the bottom of the cup.

For the turbulence jet case, the experimental results showed that jet stream started to break up mid-way between the nozzle tip and the bottom of the cup. I have used rasInterFoam, but, delta t quickly became so small that it becomes un-pratical to continue.

For both case, I applied axi-symmetric. For the rasInterFoam case, I clustered more elements near the wall and the expected pathway of the liquid/air interface. So, for these area with fine element sizes, the aspect ratio may be large. If I used uniform grid elements, then, rasInterFoam diverges very quickly.

So, the first question is: what is the difference between rasInterFoam (Reynold Averaging?) and lesInterFoam (Large Eddy)? Which one will be more suitable for this problem?

Uniform mesh in both X-dir and Y-dir does not seem to work. However, if more grid points are clustered near the expected liquid/air interface, then, aspect ratio will be high for some areas. What should I do?

Your suggestion will be highly appreciated.

Pei

[pierre]

My understanding is that RasInterFoam averages/ filters out most fluctuations, so that in the case of a jet break-up where small perturbations and waves play an important part in the physics of the break-up and where ligaments of relatively small size should appear(not just at high We), RasInterFoam won't be able to predict the physics correctly, however LesInterFoam has more chance to capture the lengthscales linked with the jet instability and dispersed structure.(mesh size important).
Using an axisymmetric geometry is ill-advised when you know that 3D turbulent and interfacial
structures have to be resolved.
What are your jet Re and We?

Pierre

[hsieh]

Hi, Pierre,

Thanks for the suggestion. I am running my case using lesInterFoam now (just in time for OpenFOAM 1.3). So far as good. Actually, there is a tutorial case for lesInterFoam called nozzleFlow2D, which is very similar to my problem.

We # is 500, Re # based on nozzle diameter is about 4,000, Re # based on the location of the start of jet breaking up is about 40,000.

I am not a "turbulence" person. Can you tell me where to find the information implemented in the tutorial case, such as:
1. how is k calculated?
2. what are c0/c1/refineCells in the case?
3. what are ptrace/Ubulk in the 0/data director?
4. what refineMeshDict is for?
5. what cellSetDict/cellSetDuct,1/cellSetDict.2 for?
6. do I need to worry about B/nuSgs/nuTilda in the 0/ directory?

Thanks!

Pei