Turbulence generation and manipulation

For nearly 100 years, the paradigm flow used for experimental investigation of turbulence has been the one generated by a regular grid moving relative to the fluid. Scientists have studied grid turbulence extensively because a properly designed grid generates turbulence that is nearly homogeneous and isotropic, and this is the idealized model assumed in most turbulence theory. Moreover, an important technical reason is that grid turbulence is often coupled with a large mean flow, particularly in a wind tunnel. This mean flow makes Eulerian measurements possible using a single probe fixed in the laboratory frame, and using Taylor's frozen turbulence hypothesis.

In the past decades, the rise of interest in the Lagrangian statistics of turbulence promoted the rapid development of Lagrangian measurement techniques. These techniques almost exclusively involve seeding the flow with tracer particles and tracking the motion of the particles. The large mean flow in a wind tunnel, however, sweeps particles out of the measurement volume in a very short time unless external mechanisms are implemented to move the measurement volume along with the mean flow. For this reason, most Lagrangian data available today are obtained from confined systems in order to be able to observe individual particles for longer times. The need to observe turbulence in such confined systems is driving research into new means to generate turbulence.

We argue below that the effort to find new ways to generate and manipulate turbulence is also inspired by the persistent need to understand flows characterized by large values of the Reynolds number, and to understand the influence of large-scale non-uniformities of the flow on the properties of the turbulence. Not only are flows with these properties seen in industrial and geophysical settings, but they are fundamental interesting because it is thought that certain statistics of the flow are only predictable in the limit of large Reynolds numbers. We discuss our work in exploring such flows in this section.

The von Kármán flow between counter-rotating disks is one of the mostly widely used means and is also one of the flows studied in our group. This turbulence can reach a very high Reynolds number, but it is neither homogeneous nor isotropic. While studying this flow, we were intrigued by two experimental findings that might also bear theoretical importance, and we discuss each of these below.

To answer these questions, we work to generate turbulence at unprecedentedly high Reynolds numbers (Rλ ~ 104) and to be able to control the degree of homogeneity and isotropy of the turbulence. It is known that an approximately homogeneous and isotropic flow with a low mean component can be achieved by using multiple loudspeaker driven jets to drive the fluid from different angles. Inspired by this work, we have constructed several apparatuses. These include a device name the Lagrangian Exploration Module (LEM), which is shaped as an icosahedron with propellers at the 12 vertices. The LEM was built in collaboration with scientists at ENS-Lyon, France. In addition, we have built loudspeaker driven flows in containers with cubic and soccerball-like geometries. These apparatuses generate turbulence in air at Reynolds numbers up to about 500, with Kolmogorov time and length scales resolvable by current measurement techniques. The isotropy of these flows can be controlled by varying the relative amplitude of the various mixers over the surface of the containers. Finally, flows with even higher Reynolds numbers (Rλ up to 104) can be generated by active grids or fractal grids in the high-pressure SF6 wind tunnel currently under construction. The active grid we are building will be the only one in which each flap in the grid is independently controllable. In this endeavor, we are collaborating with the Warhaft group at Cornell University and the Vassilicos group at Imperial College London on the design of active and fractal grids.

Contact: Eberhard Bodenschatz