Understanding differences in pipe flow between master and next

[mattfalcone1997]

I have been running turbulent pipe flow simulations in nekRS and have encountered some discrepancies between the master and next branches of nekRS. It seems that the master branch over-predicts the wall shear stress (and undershoots the log law) but the next branch seems to behave as expected. Note that such behaviour is not observed in turbulent channel flow. I've attached a plot below of the turbPipePeriodic example in both branches. I have also observed this difference in my own developing pipe flow case.

I am just wondering what the differences between the two branches are that would cause this difference and whether it may affect other flows.

[stgeke]

This is probably due to the different mesh.

[mattfalcone1997]

I just diffed the meshes and they are the same. My developing pipe flow case also uses the same mesh in both cases.

[stgeke]

I suggest carefully comparing the logfiles to identify where the differences are.