CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   LETOTs in LES (

Mana May 1, 2013 03:32

I am simulating an open channel flow in Ansys-CFX using LES model.I have gone through many papers regarding this. In almost all papers i have read this "The Ansys-CFX simulation conforms to 100 LETOTs to obtain a steady state solution and a further 25 LETOTs to gather data". what does LETOTs mean? and what is LETOTs to do with initial time period? where are the options in Ansys-CFX to incorporate LETOT values.

Thank You in advance

ghorrocks May 1, 2013 08:04

I have no idea what LETOT means. It is not a CFX term to my knowledge, the paper should define it as I suspect it is their term.

oj.bulmer May 1, 2013 08:47

LETOT= Large eddy turn-over time, ie the time the largest eddy requires to finish one rotation. From your description it appears that after period of 100 LETOTs the solution converged to steady state. Then they ran it further for 25 LETOTs to obtain the (time averaged) steady state data, just to be statistically fair etc.

You just need to find out the timescales of the largest eddies and then use its multiples to determine how long you need to run the transient simulation.


Mana May 1, 2013 15:41

thank you sir for quick reply. The time step size is based on the Large Eddy Turnover Time (LETOT) which is generally evaluated from the kolmogorov time scale. But the kolmogorov microscales has little relevance to a LES simulation (Is it so??). I am using transient solution for a time step size 0.001 s (guess value). I exactly want to know, whether it is required to incorporate the LETOTs in ANSYS -CFX. Simulation is currently running smoothly with 850 runs.
what is the meaning of natural timescale of large eddy= H/u*, and how it is related tothe length of simulation

oj.bulmer May 1, 2013 15:58

Your statements seem a bit confusing. Kolmogorov's lengthscales are considered only when you are performing a DNS. In LES you would only resolve the larger lengthscales (comparable to Taylor's lengthscales), while smaller lengthscales will be modeled. I don't exactly understand how you'd correlate LETOT with Kolmogorov eddy timescales. Two are separated by quite a margin.

I think time equal to multiple of LETOT just provides you with a measure of how long you would run the transient simulation, by thumbrule. I frankly don't know what the multiple would be, you can do a bit more literature search on this.

I guess, the natural timescale of large eddy can be roughly evaluated as the eddy size divided by the velocity scale of the eddy, the way you defined.


All times are GMT -4. The time now is 12:00.