[Nek5000-users] Finite-amplitude disturbance based on os7000

[nek5000-users at lists.mcs.anl.gov]

As a follow-up, I recompiled with lx1=18, with a dt of 1/2 the size, and IFCHAR=F, for which you will find much closer agreement with linear theory.


I have not explored all the possible parameter options - e.g., changing just dt, changing just lx1, changing just IFCHAR..., but it's clear that the default settings are not at the limit of what we can realize with the code.


hth,


Paul


________________________________
From: Nek5000-users <nek5000-users-bounces at lists.mcs.anl.gov> on behalf of nek5000-users at lists.mcs.anl.gov <nek5000-users at lists.mcs.anl.gov>
Sent: Friday, April 27, 2018 6:50:55 AM
To: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Finite-amplitude disturbance based on os7000



Dear Emily,


There could be several reasons for the discrepancy.

It could be a question of spatial and/or temporal resolution, or it could be because these amplitudes are finite, whereas the literature values are for the infinitesimal case.


This is a challenging problem on the nek side because there is a balance between having a very small perturbation - to be like linear theory - and having significant signal to overcome noise, which is at 1.e-16 + a bit more because of the iterative solver tolerances.


I tried to pick a value of the perturbation amplitude that would allow us to be as close as possible to linear theory, as measured in the growth rate.   It is certainly true that Nek will evolve the SEM solution, and not the analytical eigenfunction.


Hope this helps, and will be curious about what you find out.


Thanks!


Paul


________________________________
From: Nek5000-users <nek5000-users-bounces at lists.mcs.anl.gov> on behalf of nek5000-users at lists.mcs.anl.gov <nek5000-users at lists.mcs.anl.gov>
Sent: Thursday, April 26, 2018 11:07:02 PM
To: nek5000-users at lists.mcs.anl.gov
Subject: [Nek5000-users] Finite-amplitude disturbance based on os7000


Dear Nek users,


I run into a problem when I tried to realize a 2D finite-amplitude disturbance in the plane poiseuille flow (based on example of os7000). The code was compiled successfully and the initial conditions (base + disturbance) were all good (I verified the profiles with the literature). Both of the base flow (u = 1-y^2) and perturbation are set in the useric subroutine.  Periodic boundary condition is employed. one wavelength is simulated.


However, after the simulation starts, the perturbations somehow do not really evolve correctly; the wave shape along the streamwise was somehow distorted especially for the first few time steps (both ux' and uy'). After that, it recovered a bit but never went back to the perfect shape same as that in literature at later time steps. This issue yields some significant discrepancy with the available literature paper. I wonder if this could be due to the non-linear effect. Could anyone help me on this please?


Regards,


Emily
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20180427/b9a5e277/attachment-0001.html>