I am investigating the flow field through turbomachinery with complex geometry with an incompressible NS solver (pressure correction scheme). I have the problem, that the space between the trailing edge and the domain outflow border is quite narrow. This is due to a coupling algorithm which is applied to connect two stages of the machine in a stationary manner. So the distance between the leading edge of the first stage and the trailing edge of the second is narrow. The trailing edge has been cut off, that the flow sees a blunt body immediately in front of the outflow. Now the question arises what boundary conditions are suitable at the domain outflow. The narrow inflow area does not seem to be critical, but at the outflow the solver has serious problems. At the moment I have implemented two kinds of BoCos at the outflow: - dirlichet Pressure bocos (i.e. the pressure is precribed and the pressure correction is set to zero) - Neuman bocos (velocities are extrapolated and scaled in a way that they match the inflow mass flux. Prssure correction is extrapolated at the outflow.)

Both Bocos work well for a long area between trailing edge and outflow, but do not work at all for my problem. I am grateful for any hint! Thank you very much.

(1). I am not sure about your description of the machine. But it looks like that you are talking about the distance from the trailing edge of a blade (either a stator or a rotor) to the exit boundary of this blade row. The exit boundary of this row can be the inlet boundary of the next blade row (either a rotor or a stator), or just the eixt of the machine. (2). When a real transient flow calculation is carried out for the stage calculation (a stator and a rotor), the flow field will handle this tight spacing effect by itself. And the wake created by the blunt trailing edge of the stator will flow into the rotor. (but I am sure that you are not calculating the true transient of the whole stage problem). (3). If you are talking about the same gap between the stator and the rotor, the tight space can create a problem unless the blade turning angle ( or stagger angle) is small. Otherwise, the non-uniform flow behind the blade trailing edge plane will interact with the domain exit plane ( or the mixing plane between the stator and the rotor ) in the steady-state calculation. (4). If the code can handle this, then it is all right because this is the limit of the steady-state stage calculation. (5). Otherwise, it is a good idea to compute the flow field row by row. In this way, the inlet domain station and the exit domain station can be positioned further away from the actual machine position. If you position the domain exit condition at one blade chord length downstream from the blade trailing edge, normally it would solve the problem. The same thing is true for the domain inlet condition. There will be some interaction when the inlet is positioned too close to the blade leading edge. (6). So, in the CFD simulation of a stage calculation, the gap between the stator and the rotor and be varied in order to obtain a solution, but then the solution in the wake region will be off. And this has to be handled separately. You can't have everything. (7). The answer is you can increase the spacing between the stator and the rotor in the simulation. And if you take the row by row approach, you have more control about picking the inlet condition from the wake region of the extended domain. (8). If this is not your problem, then you will have to describe it more clearly.

like john i'm not quite sure exactly what you want. however if you'd like to place your outflow boundary near to the trailing edge (or in front of it) you might try a non-reflecting bc. now i don't know much about the types of bc usable with pressure correction methods but it you can use nonreflecting bc then they should help. you have to implement them very cleanly though. also you can take the route of solving with the outflow boundary far from the trailing edge and then just "chopping off" the domain where you want it. it's kinda crude but it should get the job done if you don't have other bc available to you.

Thank you very much for your hints. What I did, is to implement nonreflecting Bocos for the velocity components (they are extrapolated at the outlet). For the Pressure correction, I tried to use nonreflecting Bocos, then I had to scale the extrapolated velocities at the outlet to the Inlet mass flux. The other possibility I tried is to use reflecting bocos for the pressure correction, namely to set the pressure correction zero at the outlet and to prescribe the pressure. Both Bocos work well if the outlet is far from the trailing edge, but do not work well if it is close to it. I suppose, as Mr. Chien suggested, that this problem must be treated in a transient way. On the other hand we want fast solutions for industrial applications. So what I am going to do is to use overlapping grids at the coupling location in order to place the outlet further downstream.