[t_clark]

Hi all,

I'd like some perspectives on set-up of an optimisation problem using adjointShapeOptimisationFoam.

I have a turbine (extracting power from the flow) in a duct. I'm interested in optimising the duct to extract maximum power (for a given overall size) - so can represent the rotor with a momentum sink or porous zone at this stage (no need for MRF complexity).

The power varies with the mass flow through the rotor zone, and with pressure drop across that zone (Power = Force.Velocity).

However, the mass flow and the pressure drop are related; for a given duct geometry, mass flow decreases as pressure drop (thrust on the turbine) increases. There is therefore an optimal pressure drop for a given geometry.

Conversely, for a given pressure drop one should be able to maximise mass flow (therefore power) by changing the geometry.

The latter is what I'd like to do. So, the question(s) is(are):

1. Can I set up a cylindrical volume with a constant porosity, surrounded by the variable-porosity region used in adjointSOF?

2. Or, should I set up the inlet to my rotor region as an outlet (and the outlet of rotor as inlet to the main volume) and couple them together with a pressure drop to represent the rotor?

In 1. or 2. I'd then optimise mass flow to maximise power, but I'd have to do several optimisations for a range of pressure drops. So...

3. How trivial is it to maximise power take-off instead of mass flow? A matter of modifying the solver a bit, or a matter of completely rewriting the equation set?

Any perspectives or alternative approaches are very welcome.

Best regards

Tom Clark