[Nek5000-users] Convergence problem in 2D square cylinder

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

Dear Nek users,

I am a PhD student at School of Aeronautics and Astronautics, Technical 
University of Madrid, Spain.
I am really interested in using Nek5000 in order to generate a flow 
field database. After that, I will apply reduced order modelling 
techniques. The final goal is to study the reduced order models and to 
investigate possible ways to speed up the calculations.

My first run in Nek5000 is bidimensional flow over a square cylinder for 
low Reynolds numbers (Re = 100). Sadly, I am getting some convergence 
problems, Courant number grows up and explodes in a few iterations. 
Plotting the velocity fields, I have noticed that the velocity near the 
square cylinder corners is very big compared to what should be expected.

Anyone has any recommendation? Is there any way to introduce artificial 
viscosity or to increment dissipation?

Regards,

VBeltranM

-- 
Víctor Beltrán Martínez
Ph.D. Student in Applied Mathematics
Escuela Técnica Superior de Ingeniería Aeronáutica y del Espacio
Universidad Politécnica de Madrid