Search NASA⌕ Search

Engineering topics

Bisset, D. K.

Publications and source records attributed to Bisset, D. K..

Aspects of Turbulent / Non-Turbulent Interfaces

A distinct boundary between turbulent and non-turbulent regions in a fluid of otherwise constant properties is found in many laboratory and engineering turbulent flows, including jets, mixing layers, boundary layers and wakes. Generally, the flow has mean shear in at least one direction within t he turbulent zone, but the non-turbulent zones have no shear (adjacent laminar shear is a different case, e.g. transition in a boundary layer). There may be purely passive differences between the turbulent and non-turbulent zones, e.g. small variations in temperature or scalar concentration, for which turbulent mixing is an important issue. The boundary has several major characteristics of interest for the present study. Firstly, the boundary advances into the non-turbulent fluid, or in other words, nonturbulent fluid is entrained. Secondly, the change in turbulence properties across the boundary is remarkably abrupt; strong turbulent motions come close to the nonturbulent fluid, promoting entrainment. Thirdly, the boundary is irregular with a continually changing convoluted shape, which produces statistical intermittency. Its shape is contorted at all scales of the turbulent motion.

Bisset, D. K.↗

Comparison of AGE and Spectral Methods for the Simulation of Far-Wakes

Turbulent flow simulation methods based on finite differences are attractive for their simplicity, flexibility and efficiency, but not always for accuracy or stability. This report demonstrates that a good compromise is possible with the Advected Grid Explicit (AGE) method. AGE has proven to be both efficient and accurate for simulating turbulent free-shear flows, including planar mixing layers and planar jets. Its efficiency results from its localized fully explicit finite difference formulation (Bisset 1998a,b) that is very straightforward to compute, outweighing the need for a fairly small timestep. Also, most of the successful simulations were slightly under-resolved, and therefore they were, in effect, large-eddy simulations (LES) without a sub-grid-scale (SGS) model, rather than direct numerical simulations (DNS). The principle is that the role of the smallest scales of turbulent motion (when the Reynolds number is not too low) is to dissipate turbulent energy, and therefore they do not have to be simulated when the numerical method is inherently dissipative at its resolution limits. Such simulations are termed 'auto-LES' (LES with automatic SGS modeling) in this report.

Bisset, D. K.↗

An eddy viscosity calculation method for a turbulent duct flow

The mean velocity profile across a fully developed turbulent duct flow is obtained from an eddy viscosity relation combined with an empirical outer region wake function. Results are in good agreement with experiments and with direct numerical simulations in the same flow at two Reynolds numbers. In particular, the near-wall trend of the Reynolds shear stress and its variation with Reynolds number are similar to those of the simulations. The eddy viscosity method is more accurate than previous mixing length or implicit function methods.

Antonia, R. A.↗