Our research interests center around phenomena associated with turbulent flows, and flows containing more than one phase, which play an important role in industrial and environmental problems. For example, global climate changes in response to man's activities depend crucially on gas, heat, and water vapour fluxes at the atmosphere-ocean interface. These, in turn, are governed by turbulence structure near the ocean surface. Similarly, the evolution of discharges, whether routine or accidental, of hazardous materials from process and power plants is largely determined by the turbulent motion in the atmospheric boundary layer that controls dispersion, deposition, and reentrainment.

The reason vital issues such as these have remained unresolved relates to the nature of turbulence itself. Turbulence at first sight appears to be random motions comprising constantly changing eddies of various sizes and shapes. However, if the motion was truly random its description would be relatively easy--at least in statistical terms. What makes turbulence prediction such a formidable proposition is that turbulence is somewhere in between--neither completely ordered nor completely random. Recent discoveries suggest that many characteristics of turbulence depend on the presence of persistent eddying motions--sometimes called "coherent structures"--that dominate the flow dynamics and transport processes (see reference [1]).

Two parallel approaches are being taken, both of which take advantage of recent developments in large-scale computation. In the first, optical techniques are used to probe the evolution of coherent structures in laboratory experiments. These include high-speed videography (1000 frames/second) using multiple cameras to track tracer particles like microbubbles, followed by digital image processing to reconstruct the flow [2]. Another technique, often used in conjunction with the first, is 3-dimensional laser-Doppler anemometry in which the Doppler shift in light scattered from the focal point of several laser beams is detected. From this the instantaneous fluid velocity components can be unfolded. The technique may be used very close (~50 microns) to boundaries and liquid surfaces [3]. Again, digital processing of data allows on-line detection of coherent structures and enables conditional sampling of the signals to elucidate the velocity field within them.

In the second approach, direct numerical simulation of the experiments are done using either pseudospectral or high-order finite difference techniques [4]. The full set of conservation equations are solved without simplifying approximations--requiring substantial computational resources since up to 6 million modes may have to be used at every time step. These simulations can then facilitate the development of simplified models for practical use.

Several discoveries have been made using these approaches. It has been found that the primary determinant of structure is the sharpness of the mean velocity change (the shear rate) near the boundary [4] and qualitative similarities are observed whether the boundary is with a solid or with another fluid--a surprising result since eddies can attach at a fluid-fluid boundary such as the ocean surface, but cannot at a fluid-solid boundary such as at the ocean floor. At low shear rates the near-boundary structures are "patchy," being caused by impinging eddies that flatten into a pancake shape as they approach the boundary. However, at high shear rates the structures form low-speed streaks with alternating high- and low-speed regions. The low-speed streaks are periodically disrupted by spectacular instabilities called "ejections." Such instabilities have been observed before near solid boundaries but never before near the free surface between a gas and a liquid [5,6].

Another discovery is that interphase scalar fluxes, e.g. mass and heat, are almost entirely determined by the behavior of the near-boundary coherent structures alluded to above [7]. The rates are, therefore, quite different at low shears and at high shears. A theory has been developed for prediction of the transport rates and also for the transition between streaky and patchy structures [7,8,9].

A third discovery arose from studies of particle deposition and reentrainment at fluid-solid boundaries--a problem of some interest for environmental applications [10,11,12,13,14]. It was found that particles could segregate in turbulent flows--rather unexpected since turbulence is supposed to promote mixing. However, the persistent vortices associated with coherent structures tend to throw heavier-than-fluid particles out to their peripheries, whereas lighter-than-fluid particles, e.g. bubbles, go to vortex axes. Because particles can accumulate in certain regions of the flow, they can have feedback effects that profoundly change the flow even when the particle (or bubble) concentrations are low. In particular, particles can go to regions where scalar transfer rates are low and hence enhance the rates without increasing friction substantially. In this sense, they act as "smart roughness [15]."

[a]S. Banerjee, 1994 "Upwellings, Downdrafts, Whirlpools: Dominant Structures in Free-Surface Turbulence," App. Mech. Reviews 47, 5166-5172.

[b]S. Banerjee, 1995 "Structure and Transport Processes in Free-Surface Turbulence" in Two-Phase Flow Modelling and Experimentation, ed. G.P. Celata and R.K. Shah, Vol. 1, pp. 13-22, Edizioni ETS, Pisa (Plenary Lecture).

[c]Y. Pan and S. Banerjee, 1995 "Numerical Investigation of Free-Surface Turbulence in Open-Channel Flows," Phys. Fluids 7, 1649-1664.

[d]M.V. Salvetti and S. Banerjee, 1995 "A priori tests of a new dynamic subgrid-scale model for finite-difference large-eddy simulations," Phys. Fluids 7, 2831-2847.

[e]S. Banerjee, R. Martini and M. Pattison, 1995, "CLOUD - A vapour/aerosol dispersion model accounting for plume 3-D motion and heat and mass transfer between phases", J. Hazardous Materials , 46 , 231-240.

[f] A. Soldati, P. Andreussi and S. Banerjee, 1993 "Direct Numerical Simulation of P article and Fluid Motion in Electrostatic Precipitators" AIChE J. 39, 1910-1919.

[g]O.J. Nydal and S. Banerjee, 1995 "Dynamic Slug Tracking Simulations for Gas-Liquid Flow in Pipelines", Chem. Eng. Comm. (Dukler Memorial Volume) 141-142, 13-39.