[hussein93]

Hey together,
I use the solution of the steady simulation for my transient simulation, so it is my initial condition.
At the inlet, I consider the total pressure, which changes over the azimuth angle. As far as I know, the boundary condition in the steady-state simulation must be the average. As a small example with simple values. At an azimuth angle of 0 degrees I have 1 bar, at 90 degrees 2 bar, at 180 degrees 3 bar and at 360 degrees 4 bar. I consider this boundary condition in the transient simulation, which also changes with time. It is in itself a shock wave. In the stationary simulation, I now form the average for this. This would therefore be 2.5 bar, which is constant over the azimuth angle, correct? Or in other words, is it necessary to use thew average values in my steady simulation, it somehow makes no sense.

greetings

[FMDenaro]

Originally Posted by hussein93

Hey together,
I use the solution of the steady simulation for my transient simulation, so it is my initial condition.
At the inlet, I consider the total pressure, which changes over the azimuth angle. As far as I know, the boundary condition in the steady-state simulation must be the average. As a small example with simple values. At an azimuth angle of 0 degrees I have 1 bar, at 90 degrees 2 bar, at 180 degrees 3 bar and at 360 degrees 4 bar. I consider this boundary condition in the transient simulation, which also changes with time. It is in itself a shock wave. In the stationary simulation, I now form the average for this. This would therefore be 2.5 bar, which is constant over the azimuth angle, correct? Or in other words, is it necessary to use thew average values in my steady simulation, it somehow makes no sense.

greetings

It is mandatory to use unsteady Bcs if you want a physical meaningfull solution

[sbaffini]

I don't think I understand what your specific case is but, independently from that, if you are doing an unsteady simulation there are two possible scenarios:

1) (You expect that) The case has a periodic pattern, maybe because there are periodic bcs, and that is your only interest. Or similarly (yet it is completely different), the case is statistically steady but you are doing an LES or DNS, which are inherently unsteady. In both cases (with due differences) you are not really interested in the initial condition as long as it doesn't prevent you to reach your periodic or statistically steady regime. In this scenario, the closer the initial condition to the final regime the better, but there's a lot of freedom.

2) All the other cases where an unsteady simulation is required. In this case, by definition, you are interested in the whole story, from beginning to end. Which means you can't be that approximate on your initial conditions.