Learning Forum

[kevin.pancardoa]

Hello everyone, as part of my college project I need to replicate the results from the paper https://www.mdpi.com/2076-3417/8/4/574 which involves a conical nozzle, however when I use the same boundary conditions I am not getting the same results and I get a marging of almost 50% of error.

As shown in the image, it is a 2D and symmetrical analysis, the paper uses an inlet of 1e06 Pa (also I am assuming 0 Pa at the supersonic/initial gauge pressure, I do not know if here is the issue), and pressure-far-field conditions of 11.1 kPa and Mach number of 2.5, the symmetry as axis and wall as wall with no slip condition. The solver is density based, steady and axisymmetric. The mesh in the paper uses 0.01m of element size with refinement in the inner walls of the nozze and they have around 120,000 elements, but I'm using 10mm of element size (0.01m as well) having a total of 300,000 elements. Also I am using residuals of 1e-06. I am also assuming the operating pressure to  0 Pa since they assume a flight altitude of 50,000ft and mach number of 2.5 (I am attaching the boundary conditions from the paper in the last image). I also suspect the problem is related with operating pressure as well as the presurre in the pressure farfield condition of 11.1kPa.

My concern is that they use k-omega turbulence model (it is not specified if they used SST) but when I use k-omega SST, I set 4,500 iterations, my simulation is not even converging, I get residuals of 1e12 leading to wrong results. However I tried using the k-epsilon model with the "realizable" option and it is still not converging but I get more realistics results, however with a high marging of error with the maximum mach number of 7.0 as shown on the second picture but I am suppposed to get a maximum mach number of 4.40 as shown in the third image. 

What do you think guys can be the issue or how can I fix it? I would really appreciate your help.

 

[Essence]

Hello,

How is the mesh? How have initialized the case?

[kevin.pancardoa]

This is how the mesh looks like, it has 341,000 elements of 10mm size. I used Hybrid initialization.

[Essence]

I mean, what is the mesh quality (Ortho. quality, aspect ratio, growth rate). Please use FMG.

[kevin.pancardoa]

Hello, finally I managed to converge the simulation. I changed the courant number from the default value of 8 to 1, now it's converging and I get really good results. Thank you for the help anyways.

[Essence]

Glad to hear that.