Discussion Forum

I'm working on a pressure exchanger model using rotating machinery, fluid flow interface with k-e turbulence model (frozen rotor). It consists of a rotor with a number of ducts exposed to stationary inlet and outlet ports. The duct radius is 10 mm with 130 mm height and the rotor rotates at 150 rpm about its axis, the port's height is 50mm. I used one duct with 2 ports for illustration. I used a velocity inlet boundary condition at the lower boundary of the lower port (.8 m/s), a pressure outlet one at the upper boundary of the upper port, and an identity pair (flow continuity boundary condition with wall fallback feature) at the boundaries in contact. The image attached below. when I set the outlet pressure to 0 Pa, the results seem logical, and the max velocity is about 3.4 m/s. when I change the pressure from 0 Pa (increasing or decreasing), the velocity at the contact between the duct and port jumps dramatically ( 52 m/s at 6MPa outlet pressure for example) and that causes flow disturbance in the ports and the duct. The velocity field should not change with outlet pressure change, but the result does not seem logical. That issue eliminates only when I use the (create imprints) option while forming an assembly. With the Imprints enabled the mesh is still non-conforming but the edges of the mating boundaries are respected. Although the nodes on either side of the Pair will not be shared, the meshes on either side will both have nodes along the mating edges. It's practical when using frozen rotor simulation, but I intend to use the frozen rotor as an initial solution of a time-dependent one and that solution becomes impractical. The weird thing is that the issue appears when I change the outlet pressure from 0 Pa. Can anyone help me solving this issue? Model file and result images attached below