jwely / pivpr Goto Github PK

only the bottom and right quadrants have fluctuating components for uu and vv respectively, and this leads to some very strange things in the turbulent kinetic energy calculation.

this appears to be related to the switch over to an rms based averaging scheme. investigate

Uncertainty analysis is showing an exceptionally strong bias in the Z velocity measurements, which gets better as the interrogation plane moves down stream.

This bias can be checked for by examining the mean axial velocity measured in each vortex compared to the free stream velocity. In every single case, the mean axial velocity is slightly elevated from the free stream velocity measured with dynamic pressure, precisely as we would expect. This seems to indicate that in reality, Z measurements are not significantly biased.

One attempted fix is to remove the meshgrid recalculation for a small z displacement in the ArtificialPIV instance which generates the image pairs. We will see how it goes, but must wait yet many more hours for it to finish computing.

the way dynamic data is tossed out before pickling creates problems if the core has not already been found prior to pickling. This was probably introduced when the core finding method was refined to iteratively solve for the core location while slowly narrowing the search field if characteristics were outside the reasonable range.

Hi @Jwely

the repo popped up again in my github page and I recall that we wanted to discuss some option to merge your ideas into openpiv-python. Thanks for mentioning us on the README, we did improved quite a lot since then. So, maybe we shall talk?

Alex @alexlib

test!