Simulation not moving from "Timestep: 1"

[danrhouck]

I have a new simulation that is not moving on from printing "Timestep: 1" after it loads the geometry and generates matrices. I didn't note my start time, but I'd say it's been there at least an hour whereas, typically, there's barely a pause there if any delay. It had more blade elements than I had tried before, so I arbitrarily halved them and tried again, but it seems the same so far. I also tried another server in case it was hung up or something, but again no change. What could cause this?

UPDATE:
A colleague advised that I troubleshoot this by removing one polar at a time and rebuild the geometry without it. After removing the polar for the cicular foil at the root, the simulation runs. I have, however, successfully used the exact same polar for a circular foil with other blade geometries, so it's not clear to me why it caused a problem here. Just in case, I added additional Reynolds numbers to the polar file to see if that could have been the issue. Upon trying to run again with the new polar for the circular foil, I got the error Warning: NaN encountered in wall_ind_vel().

So now my general question is what are some recommended or best practices for building the root sections of a blade geometry? There's an additional problem in that tools like XFOIL have difficulty producing polars for thick airfoils that transition from the cicular root to the first "true" airfoil section. Without the cicular airfoil, the geometry I'm working on now begins at r/R of 0.12 because I cannot generate any polars closer in. On the other hand, with the cicular airfoil, its polar is used for the first four of 48 elements (up until the next available polar), which also doesn't seem very accurate. Any suggestions?

[whophil]

So now my general question is what are some recommended or best practices for building the root sections of a blade geometry? There's an additional problem in that tools like XFOIL have difficulty producing polars for thick airfoils that transition from the cicular root to the first "true" airfoil section. Without the cicular airfoil, the geometry I'm working on now begins at r/R of 0.12 because I cannot generate any polars closer in. On the other hand, with the cicular airfoil, its polar is used for the first four of 48 elements (up until the next available polar), which also doesn't seem very accurate. Any suggestions?

In the absence of wind tunnel data, polars for thick airfoils can be interpolated. But for typical blades which transition to a circle, the lift of airfoils very far inboard will likely have a small effect on any quantities of interest (in vortex lattice models like CACTUS). You might try a pure interpolation to get from the airfoil data you have at r/R = 0.12 to an analytical or empirical polar at the cylinder. You can use the blending capability from https://github.com/WISDEM/AirfoilPreppy for this.