Universality and scaling relations for turbulent/non-turbulent interfaces in free shear flows
Please login to view abstract download link
The wake of a bus, the jet from an airplane engine, and the air current inside a room are just three examples of turbulent free shear flows whose consequences can be easily observed. In these and many other engineering and natural flows the central turbulent stream is bounded by fluid in a non-turbulent state, and the two regions are separated by a sharp and highly convoluted interface: the so-called turbulent/non-turbulent interface layer (TNTI). Through this interface important exchanges of mass, momentum, and scalars take place, in a process through which non-turbulent fluid becomes turbulent (turbulent entrainment), which is strongly governed by the TNTI characteristics. Although this processes is present in many important turbulent flows, many important aspects of this mechanism have not been fully clarified. In my PhD thesis some of these questions have been investigated by using direct numerical simulation (DNS) of jets and wakes. In DNS all the relevant scales of motion are explicitly simulated from the large energy containing scales of motion associated with the kinetic energy of the flow, down to the smallest scale motions associated with the energy dissipation. DNS is virtually unsurpassed by any other simulation technique and is equivalent to an experiment, and permits the access to the full unsteady, three-dimensional, multiscale complexity of any flow quantities of interest. During my PhD thesis I have carried out DNS which I used latter to undertake a comprehensive and thorough investigation of the TNTI, where I developed some of the numerical tools that made this investigation possible. One of the most significant outcomes from these tools is the ability to obtain conditional statistics for any flow variable, as function the distance from the TNTI. The present results obtained allowed to establish the detailed geometrical characteristics TNTI layer, and in particular its thickness of the TNTI at high Reynolds numbers. I was also able to show that, when properly normalised, statistics of the main flow variables in the vicinity of the TNTI layer are universal. This result opens the door to the possibility of developing simple engineering estimates of some flow mechanisms near TNTIs, for routine engineering computations.