[Cnl5]

Dear all,

Currently, I am using wall-resolving implicit large eddy simulation to simulate turbulent channel flow. The channel is periodic in streamwise and spanwise direction while a non-slip condition is applied at the bottom wall. The calculated y+ is below 1. However, the velocity profile from the LES is overestimated in log-region as compared to that of Spalding. Does anyone face this problem before?

https://imgur.com/a/Z2QD083

[FMDenaro]

Originally Posted by Cnl5

Dear all,

Currently, I am using wall-resolving implicit large eddy simulation to simulate turbulent channel flow. The channel is periodic in streamwise and spanwise direction while a non-slip condition is applied at the bottom wall. The calculated y+ is below 1. However, the velocity profile from the LES is overestimated in log-region as compared to that of Spalding. Does anyone face this problem before?

https://imgur.com/a/Z2QD084

Overestimation of the LES velocity profile is often reported. Have a look here https://www.researchgate.net/publica...mulation_codes


However, you profile in the centerline shows a wrong behavior and I think that something is not correct. Have you waited enough time to let the fields correlating physically from the initial conditions? Check the total kinetic energy in time. How many samples have you used for the statistical averaging?

[Avr.Tomer]

Originally Posted by FMDenaro

Overestimation of the LES velocity profile is often reported. Have a look here https://www.researchgate.net/publica...mulation_codes


However, you profile in the centerline shows a wrong behavior and I think that something is not correct. Have you waited enough time to let the fields correlating physically from the initial conditions? Check the total kinetic energy in time. How many samples have you used for the statistical averaging?

Awesome reply and ref! I have a few months of LES ahead of me. Thanks Prof. Denaro!

[FMDenaro]

Originally Posted by Avr.Tomer

Awesome reply and ref! I have a few months of LES ahead of me. Thanks Prof. Denaro!

Hello,
this overestimation effect has been observed using several discretizations and, somehow strangely, often does not appear in the no-model LES.


You can have also a look to this paper https://www.cambridge.org/core/journ...9AE603FD6BB11B