[GTP] New Developments in Turbulence Modeling. I. Theory November 4, 2010

[Carolyn Mueller]

Stefan Heinz

 

University of Wyoming, Department of Mathematics

 

For the foreseeable future we have to calculate turbulent flows as 
observed in reality on the basis of equations that involve a turbulence 
model. The development of computational methods that have a predictive 
power requires the use of equations for large eddy simulation (LES). In 
such equations we need a subgrid-scale (SGS) turbulence model that 
accounts for the interaction of large and small scales. Existing SGS 
models face three kinds of problems. First, the consistency of models is 
not always clear: a turbulence model describes correlations of a 
stochastic process, but there are SGS models for which an underlying 
stochastic process does not exist. Second, most SGS models need 
adjustments to the flow considered: such methods have a limited 
predictive power. Third, LES are often too expensive for applications to 
wall-bounded flows at high Reynolds numbers, which have to be computed 
in many applications. Theoretical solutions to these problems will be 
described in this first part of two talks by explaining the basics of 
stochastic turbulence modeling and the extension of these methods to 
dynamic and unified turbulence models. The existence of an underlying 
stochastic turbulence model overcomes the consistency issue. The dynamic 
model formulation overcomes the need for adjusting the model to the flow 
considered (this formulation does also overcome the adjustment of 
existing dynamic methods by clipping or averaging coefficients to ensure 
the stability of solutions). The unified model formulation overcomes the 
cost issue of existing LES methods by the consistent integration of LES 
and ensemble averaged equations.

November 4, 2010
FL2 Main Auditorium
Room 1022
 Lecture 3:30 pm

-- 
Carolyn Mueller
NCAR IMAGe
1850 Table Mesa Drive 
Boulder, CO 80305 
www.image.ucar.edu 
Tel: 303 497-2491
Fax: 303-497-2483