• Home
• Faculty
• Research
• Laboratories
• Contact us
• Thermosciences
• Companies

Research Projects in Experimental and Computational Fluid Mechanics

Experimental Fluid Mechanics

The equations of motion of simple "Newtonian" fluids like air and water have been known for over a century, but direct numerical solution is economical only for laminar flows, and most fluid flows in engineering are turbulent. Direct solutions for turbulent flow, even in simple geometries at low Reynolds number, can require hundreds of hours of supercomputer time. Simplified methods of predicting momentum and heat transfer (developed by turbulence modeling) need empirical information, and of course a qualitative understanding of turbulence physics, both obtained mainly from laboratory experiments. Therefore experimental fluid mechanics is alive and well, and our research group has made many significant contributions over the years. Current work is described in the PROJECT LIST under the names of the faculty members listed below.

Our experiments are done in a range of wind tunnels and water tunnels (or open channels). We have about a dozen major TEST RIGS and a number of smaller, or special-purpose, facilities. From time to time we also use the experimental facilities at NASA Ames Research Center.

The main techniques for measuring the velocity fluctuations in turbulence, which have frequencies up to many kilohertz, are the hot-wire anemometer (HWA), the laser Doppler velocimeter (LDV) and a relatively new technique called particle image velocimetry (PIV). The HWA relates velocity to rate of heat transfer from an electrically-heated wire, the LDV deduces velocity from the frequency of light reflected by particles moving through an optical fringe pattern, and PIV deduces velocity from the distance moved by illuminated particles between successive photographic images. Generally speaking, hot wires are easiest to use in air, LDVs in water.

Computational Fluid Mechanics

The physics of any fluid flow are governed by a set of partial differential equations known as the Navier-Stokes equations. These equations are highly nonlinear and cannot be solved analytically for any but the simplest flows. Computational fluid dynamics encompasses many fields which are involved with the numerical solution of these and related equations. The rapid improvement of both the speed and memory size of computers is facilitating advances in numerical computation of fluid flows. The department has a substantial program on computation of turbulent flows.

Direct Numerical Simulation (DNS) solves the unsteady Navier-Stokes equations in full, using highly refined meshes on supercomputers. Extensive computer time is required, but DNS can provide "exact" solutions to the Navier-Stokes equations. In addition, it can provide turbulence modellers with virtually any flow quantities of interest to enable them to validate and calibrate their models.

The Reynolds Averaged Navier-Stokes (RANS) equations are based on a statistical approach. Computation of turbulent flows using the Reynolds-averaged form of the Navier-Stokes equations with appropriate turbulence models is an active area of research. This is the most practical approach to computational fluid dynamics for flows of engineering interest.

Large Eddy Simulation (LES) rests on the theory that the small eddies in any turbulent flow are nearly universal, while the larger scales are geometry dependent. Therefore, the small eddies can be modelled while the effect of the larger eddies is solved directly using the Navier-Stokes equations. In this sense it is a compromise between DNS and RANS. Although not as expensive as DNS, LES still requires extensive computer resources.

Flow Physics and Computational Engineering is also very active in computational fluid dynamics.

A list of current and recent research projects in Experimental Fluid Mechanics is given here with the name of the faculty and Ph.D students involved. (The last item is the name of the sponsor.)

Prof. Peter Bradshaw