ARTICLE:

• TURBULENT SCALAR MIXING IN SINGLE AND DOUBLE ELBOW CONFIGURATIONS  download article

V. K. Vijayaraghavan
Texas A&M University, College Station, USA

Y. A. Hassan
Department of Nuclear Engineering, Texas A&M University, College Station, TX 77843, USA

A. R. McFarland
Department of Mechanical Engineering, Texas A&M University, College Station, Texas 77843

ABSTRACT

Numerical simulations were performed to study the turbulent mixing of a scalar species in single and double elbow flow configurations. The double elbow configurations studied included S and U shaped double elbows in a single plane. Different Reynolds Averaged Navier Stokes (RANS) and Large Eddy Simulation (LES) models were used to model the turbulence in the flow. The RANS models used for the simulation included standard, realizable, Renormalization Group (RNG) k-ε and the Reynolds Stress Models. Static Smagorinsky sub-grid scale model was used for the LES simulations. Wall functions were used to resolve the near wall boundary layer. These simulations were run with both two-dimensional and three-dimensional geometries. All the RANS simulations were steady simulations, while by definition the LES simulations were unsteady. The velocity and tracer gas concentration Coefficient of Variations (COVs) were compared with experimental results. The COV, defined as the ratio of the standard deviation to the mean for a set of data, is a measure for the degree of uniformity of the data. The results from the LES simulations compared better with experimental results than the results from the RANS simulations. For example, at a distance of 10 diameters downstream from the exit plane of the single elbow, the concentration COV of 29% predicted using LES is closer to the experimentally predicted value of 20% than the value of 86% predicted using RNG k-ε model. In addition, the level of mixing downstream of a S-shaped double elbow was higher than either the single elbow or the U-shaped double elbow due to the presence of counter rotating vortices. It can be concluded from the comparison of numerical results with experiments that, while LES may be an acceptable way to demonstrate compliance with standards, the present RANS models used for scalar mixing may not be acceptable.

 download article

TRB-28 pages

« Previous article         Next article »