We have a lab experiment in which we collected temperature data for flow over a heated flat plate. The local Re range from 15,000 - 350,000. Our Fluent model matches up very well with the lower range Re, but at the higher end there is quite a difference. Is there something we should change in the k-epsilon model or maybe change the model type in general as we reach the higher Re?..

thanks

The Reynolds number is a measure of the ratio of inertia forces to viscous forces. It can be used to characterize flow characteristics over a flat plate. Values under 500,000 are classified as Laminar flow where values from 500,000 to 1,000,000 are deemed Turbulent flow. It is important to distinguish between turbulent and non turbulent flow. http://kay.gemba.org/pdf/MAE440/MAE440Exp03.pdf

1- Why are you treating a laminar flow as turbulent?

or why are introducing in your calculations something that does not exist (the Reynolds stress)?

2- You mention a heated plate, do you mean isothermal plate, constant heat flux or what?

May be you should consider the Rayleigh number as your criteria

Yes agree, with P2, unless typo, you are laminar

can you post your data for review?

the problem we are having is that the Re in the square duct is far above 2400. So, we are unsure of the treatment between Turbulent or Laminar..

thanks for the link.

tis different now. duct flow Re(Dh) = 4*h * u/mu, Re.T ~ 2200 yes. Ok so you are in need of first solving the flow as fully developed, turbulent, no?

the data match problem arises at the higher Res. At the lower Res the Fluent model data matches very well to the lab experiment. But, diverges sharply at the higher Re.

The duct Re >> 2400.

The Fluent model is replica of the lab experiment. I am trying to get some suggestions as to why the Fluent model would diverge from the lab data at the higher Res.

thanks

[baumert]

Hi,

the k-Eps model is in most cases (there are many different
parameter sets on the market) a low-Re closure
which fails at solid boundaries in that it predicts a wrong
von-Karman constant. There is now a renovated K-Omega
model available which rests on a new view of turbulence
as a kind of gas of vortex dipoles. It predicts the
von-Karman constant as 1/sqrt(2 pi) = 0.399
which is close to the standard value 0.4. I recommend to use the new
renovated K-Omega (RENO) at very high Re. For details
see chapters 3, 4, 5 in the following book:
http://books.google.com/books?id=HVqbdXI29i0C&hl=de
Further, a log profile we can expect ONLY for very high Re.
At low Re we get the Blasius or Van Veen profile for the velocity.

Regards, hzb