סמינר מחלקתי Daniel M. Israel

School of Mechanical Engineering Seminar
Thursday, July 23, 2015 at 15:00
Wolfson Building of Mechanical Engineering, Room 234

Low Order Models for Assessing RANS Models

Daniel M. Israel

Historically, most coefficients in RANS models have been calibrated to match the growth rates of certain canonical self-similar flows. However, for many of these flows, the growth rates observed in experiments and DNS vary widely. In fact, George (1986) argues that a universal self- similar solution does not exist. This would imply that RANS model calibration is specific to a particular experiment. Using classical integral methods to reduce RANS models to ODEs, it is possible to obtain a low-order dynamical system which can be used to study the approach to self-similarity for the model. Comparing the trajectory maps for such low-order models to data suggests that most, if not all, of the discrepancy between different experiments can be explained by transient deviations from self-similarity, and that there is indeed a universal self-similar behavior. Furthermore, such trajectory maps can be used to assess how well transient behavior due to the initial conditions in RANS calculations captures the experimentally observed flow physics.