[fivos]

Hi to everyone,

My supervisor asked me to simulate the flow at the nozzle of a Pelton turbine including the piping before the nozzle and the jet leaving from the nozzle, in order to study possible flow non-uniformities arising from the geometry of the inlet pipe and their effect on the jet shape. I have access from my university to both Fluent & CFX.

I have already solved individually the flow at the inlet piping till section B at the attached picture "nozzle_section.jpg" (sections refer to the purple circular surfaces accross the pipe - see also the "nozzle.jpg" picture for a better understanding of the geometry it shows a section view at the plane of symmetry). The flow at the inlet piping was solved as 3D steady with k-e turbulence and the flow at the nozzle was solved as axis-symmetric unsteady using VOF again with k-e for turbulence.

Solving the whole pipe as 3D unsteady with VOF is total overkill, since the geometry will consist of lots of elements and convergence requires some time for the steady case; in the unsteady case things will be much worse.
Also I am unable to apply symmetry planes to reduce elements number, since the geometry as a whole is not symmetric (even if in nozzle_section.jpg seems so - the pipe geometry extends beyond the picture and is not symmetric).

Would it be possible to solve as 3D steady until, lets say, section B or C and then take this solution and use it as a boundary condition for the 3D unsteady VOF simulation of the nozzle (can this be considered valid)? In that case, I guess that UDFs will be needed for defining the inlet velocity profile at the inlet of the nozzle. As far as I have seen the define_profile UDF might do that. However the define_profile works nice with simple polynomial profiles. Would it be possible to express the complex velocity profile at the nozzle inlet for the VOF calculation? If yes how?

If anyone else has encountered such problems, or has any ideas on these please do not hesitate to respond.

Thanks in advance.

PS. I would like some advice on the meshing of this geometry too. The problem is that untill section A the geometry can be considered topologically a cylinder and can be easily meshed using Cooper scheme in Gambit with boundary layer. But when the nozzle with the needle is added then things become really difficult, since the geometry cannot be considered a cylinder any more and I must use Tgrid mesh. It is hard to create a good boundary layer at the tip of the needle and at the point where the needle exits the piping there are mesh quality issues.
If anyone has suggestions for meshing the geometry, he is more than welcome to mention them.