[coolcrasher]

Hi all,

Last month we have tried running the simulations with 3 different meshes and also with different timesteps. I have attached the excel sheet which has the results.

We have seen that using movingWall velocity boundary condition on the patches which are rotating give meaningful results and also we have used the same boundary conditions for others as used in propeller tutorial. ( for pressure, k, omega, nut. KomegaSST Turb model is used ).

Three different meshes are as follows
1) Basic mesh ( which has been used previously )
2) Finer mesh ( Basic Mesh is taken, mesh at all regions except AMI patches are made finer )
3) Finer AMI mesh ( Finer Mesh is taken, mesh at AMI patches are made much finer )

Following are the observations found:
1) We still observe the flowrate difference between the AMI patches.
2) Flowrate difference is less in case of Finer AMI mesh compared to Basic Mesh. That means that as we refine the AMI patches further we observe that the flowrate error is reducing for a particular timestep.
3) For a type of Mesh we observe that as we get no flowrate difference between the AMI patches at a bigger timestep ( like 1e-2 ) but as we make the time step smaller ( till 1e-5 ) we observe that the error is gradually increasing. ( which is quite strange, we expect to get better results as we take finer time steps ).

We have tried running the Basic case with time step of 1e-4 with k epsilon model. It has ran fine till first few time-steps ( around 2-3 revs of impeller ). Then the case has diverged and crashed. But even during that time we have observed the loss in flowrate.

The problem of flowrate mismatch has not been solved yet. Also we are confused what is causing this error. Is it due to bug in OpenFoam or due to some wrong boundary conditions used.

Please help me with this regard.