constant lift simulation

bennn · August 19, 2016, 05:46

Hi all,

I'm doing optimization for 2D airfoils with simpleFoam. I'd like to operate at constant lift, by updating the inflow vector. So far I tried the following approaches :

- Code an adaptative inflow version of simpleFoam, but this gets very unsteady for non classical shapes...

- run two simulations, and find the adequate lift by using a linear inter/extrapolation. This is quite accurate but requires 3 runs...

- run a low order model first - xfoil is perfect for that since it has the ability to do constant Cl - and use the angle as an input to openfoam. This is very efficient, but can be more than 10% wrong. And low order model does not converge that easily for non classical shapes. And that would be hard to apply to 3D shapes.

Any idea is welcome ! I know SU2 operates in constant lift by updating the inflow (option #1 above), and it is also a bit unstable. I think the best option might be to get the load from potentialFoam, but I couldn't find a way to do that.

Thank you all for your input !

Artur · August 23, 2016, 07:27

Hi,

My guess is you would need to be very careful with updating the inflow angle, especially at the start of the simulation, to avoid rapid changes as this would likely cause divergence. You'd probably need to have some sort of "underrelaxation" feature built-in and do the update only every couple of time steps and even that after some initial level of convergence has been reached.

Another option would be to use dynamic mesh, like in the incompressible/pimpleDyMFoam/wingMotion tutorial. This would probably require you to first run the case at an approximate AoA with simple foam, restart from the last time step with pimpleDyMFoam and use a custom motion solver which would adjust the AoA for the desired Cl with some kind of optimisation/controller routine robust enough not to cause too rapid changes.

Unfortunately I haven't dabbled in either so can only share my ideas and not actual code.

All the best,

Artur

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
How to calculate lift, drag and lift distribution in Star CCM+?	israelcasillas94	STAR-CCM+	4	November 6, 2020 08:35
decomposePar problem: Cell 0contains face labels out of range	vaina74	OpenFOAM Pre-Processing	37	July 20, 2020 05:38
Simulation and Optimisation of centrifugal fan 3D to 2D	eRzBeNgEl	STAR-CCM+	0	January 31, 2013 13:21
Boundary condition setting regarding turbine simulation using CFX	Lacerlacer	CFX	11	March 12, 2012 09:32
How obtain the average of lift over time for a transient simulation?	aero	ANSYS	0	November 11, 2009 02:00

August 19, 2016, 05:46	constant lift simulation	#1
bennn Member benoit paillard Join Date: Mar 2010 Posts: 96 Rep Power: 16	Hi all, I'm doing optimization for 2D airfoils with simpleFoam. I'd like to operate at constant lift, by updating the inflow vector. So far I tried the following approaches : - Code an adaptative inflow version of simpleFoam, but this gets very unsteady for non classical shapes... - run two simulations, and find the adequate lift by using a linear inter/extrapolation. This is quite accurate but requires 3 runs... - run a low order model first - xfoil is perfect for that since it has the ability to do constant Cl - and use the angle as an input to openfoam. This is very efficient, but can be more than 10% wrong. And low order model does not converge that easily for non classical shapes. And that would be hard to apply to 3D shapes. Any idea is welcome ! I know SU2 operates in constant lift by updating the inflow (option #1 above), and it is also a bit unstable. I think the best option might be to get the load from potentialFoam, but I couldn't find a way to do that. Thank you all for your input !

August 23, 2016, 07:27		#2
Artur Senior Member Artur Join Date: May 2013 Location: Southampton, UK Posts: 372 Rep Power: 19	Hi, My guess is you would need to be very careful with updating the inflow angle, especially at the start of the simulation, to avoid rapid changes as this would likely cause divergence. You'd probably need to have some sort of "underrelaxation" feature built-in and do the update only every couple of time steps and even that after some initial level of convergence has been reached. Another option would be to use dynamic mesh, like in the incompressible/pimpleDyMFoam/wingMotion tutorial. This would probably require you to first run the case at an approximate AoA with simple foam, restart from the last time step with pimpleDyMFoam and use a custom motion solver which would adjust the AoA for the desired Cl with some kind of optimisation/controller routine robust enough not to cause too rapid changes. Unfortunately I haven't dabbled in either so can only share my ideas and not actual code. All the best, Artur