[DA6righthand]

I'm in need of advice running a simulation. This is my first experience with LES so I'm new to the solution procedure. My project concerns a rectangular single expansion ramp nozzle which I'm only simulating half of the total domain because it's symmetric about one axis. The nozzle is being ran at a nozzle pressure ratio (ratio of nozzle inlet total pressure to free stream pressure) of 3. Initially this project was intended to be ran only as steady but the need to expand analysis to LES was deemed necessary. Since LES wasn't initially planned, my fluid domain is a bit smaller than I feel is appropriate. I used the same grid as I did for the steady simulation (~30 million elements, ~9 million nodes, good orthogonal quality and skewness). Here's what's been done thus far:

- I began running the Stress-Blended Eddy Simulation (hybrid RANS/LES) from the steady solution result.

- A time step size of s was determined from the smallest element size in the jet and velocity of the flow.

- I wanted to run the simulation out to 2 ms (4000 time steps).

After 2182 time steps (1091 microseconds flow time), the below picture was generated to check the solution. It's the schlieren image created in CFD-Post plotted on multiple transparent slice planes using an inverse logarithmic greyscale.

NPR3p0v1_9SBES_1p41275msftSchlieren by Justin Holder, on Flickr

Yes, the simulation is not yet complete as it has not yet fully developed. The full domain is pictured in this plane (above, below, and aft of the jet). As you can see, I'm getting pressure waves reflecting off the left face (a pressure-inlet boundary), the top, bottom and left end (all pressure-outlet boundaries). I attempted to apply non-reflecting boundary conditions to all external boundaries but my solution fails when I do so. My questions are:

1) At what point can I enable non-reflecting conditions?

2) When I do apply non-reflecting conditions, do I need to reduce my time step size? If so, by what amount?

3) Will I even be able to use LES with this domain (i.e., my domain is too small)?

As we speak, the simulation is continuing to run. I'm just not sure of how, if I even can, get a good solution using LES with this domain. Any help is super appreciated!

Thanks,
Justin