[Eglantin]

Hi,

I am running a 3D CFD on an axisymmetric body (a 18deg sector) of a fuel injection system. The purpose is to extract the CH4 mass fraction profile at a specific location.

When running the CFD after generating the mesh with FLUENT Meshing, I noticed unstable behavior of the solution in steady state.
Therefore I switched to transient and the unsteady behavior continued. I concluded then that the unstable flow is characteristic of the domain.

What I did next is running different transient simulation with different time-steps and perform FFT of one of my monitors. Based on the frequency, I extracted the main period of the oscillation.

This was useful since it helped me understand how many periods I wanted to average, therefore how many timesteps. I proceeded then running the simulations for certain, say, 20k timesteps and averaged only after 15k the exact number of timesteps that would correspond to 5 periods.

The question is: is this a reasonable approach? All my previous CFD had stable solutions but this new domain is giving me a hard time due to its instability.

NOTE: The timestep has also a great impact on the profile so I did a time-step sensitivity study.

A feedback would be very much appreciated.

Thanks