If this looks familiar, its because I've previously posted this on the Solidworks subreddit.
I'm modelling my final year project in Solidworks and I've run into a bit of a snag. I will gladly hold my hand up and say that I'm fairly new to flow simulation though.
I'm studying plug nozzles and I've simulated one of these mounted in a proposed test rig - I thought I'd simulate running gas through the chamber and nozzle, read off the forces for a ballpark figure of the thrust I can expect to get and then run some FEA to determine whether I should be using strain gauges to measure thrust via deformation or have the articles mounted in a spring-mounted cradle, measuring thrust against the compression of the springs.
However, despite my test geometry being symmetrical in two planes I am getting decidedly assymetrical forces when I look at surface goals.
2D CFD plot
This shows the flow has a slight bias towards the bottom of the image, but I have no idea what is causing this deflection of flow.
3D surface goals & overview
X is the axis of the assembly along the centerline. Note Y & Z values nowhere close to symmetrical - in fact I'm getting more thrust downwards and sideways as I am along the axis.
Boundary conditions are pretty minimal: http://i.imgur.com/eSt1MPU.png 
Working fluid is Nitrogen gas. I havent used the high mach flow option as for some reason regardless of geometry the flow solver simply iterates through maybe five or six iterations before massive discontinuities appear and the solver bricks itself.
Following the comments left on the Reddit thread, I went back and re-ran the simulation with Gravity turned on and tried to force the solver to take as many iterations as possible by starting off with a crude mesh (setting 2) and then setting the refinement option in calculation control options to 1. I'm still getting flow diverting to one side of the central body with a corresponding deformation of the exit plume, but this time to the *top* of the assembly not the bottom.
Am I doing something wrong here with regards to the settings and boundary conditions?
I guess my next steps are to run the simulation over only half the current computational domain with a symmetry relation, or just to try altering the model geometry to check that I've not accidentally managed to create a model that creates an unstable flow condition.