Writing lines in turbulent air using Air Photolysis and Recombination Tracking

From the 16th century, turbulence has remained a topic of continuous study, stemming both from academic and industrial interest. In this thesis a new scheme of Molecular Tagging Velocimetry (MTV) called APART (Air Photolysis And Recombination Tracking) for measuring this turbulence is introduced. In MTV a pulsed laser is used to ‘write’ a pattern of molecules in the flow field. After a set time delay the pattern, that has been altered by the flow field, is read back. The velocity field now follows from the displacement and the deformation of the pattern. In order to resolve the displacement of the pattern it is necessary to do careful image analysis. By employing a two-stage fitting technique the line center of all but the most deformed lines can be found. Although several different schemes of creating and visualising molecular tracers have been developed, APART is one of the most promising. In APART tagging is done by photosynthesis of NO molecules out of N2 and O2 molecules in air. No additional seeding of air is needed to create these molecular tracers. Furthermore, NO is a stable molecule and is thus long-lived. Although one may naively expect that, since APART employs extremely small tracers particles, that is molecules, this allows the resolution of even the smallest turbulence scales. However, we show that, since molecules reside at such different scales, they are governed by different mechanics, resulting in diffusion. For NO lines in air it is shown that, no matter how thin initially, these will broaden by thermal diffusion to the size of the Kolmogorov length within one Kolmogorov time, and will thus smear out turbulence on the smallest scale. In order to compare APART with known flow properties, fully characterised turbulence is used. It was created by means of a free turbulent jet where, at approximately 45 nozzle diameters, Re¿ ?? 460. All macro scale parameters of the flow were determined and linked to the micro scale properties by the relation between the rms velocity urms and the dissipation rate ??. Histograms of the velocity indicate that the velocity distribution is almost perfectly Gaussian for all positions along the written line, even though other line properties such as intensity and width are position dependent. Though APART is an unseeded optical technique, it can not be considered fully non-intrusive. That is, energy absorption along the path of the laser beam, mainly due to oxygen leads to local heating. Through different techniques (energy absorption, LIF spectroscopy and line width analysis) it is shown that the rise in temperature within the first microseconds is approximately 400K. Not only does the laser based heating increase the thermal diffusion coefficient, it also results in convection. These effects have been calculated and are shown to produce super-diffusive line widening. Indeed, this effect is observed in measurements. Another effect that was predicted and has been measured is that the diffusion constant D does not go to zero for infinite pressure (1/p ¿ 0). Since APART allows us to write lines and we can observe the displacement perpendicular to the line, we can measure true transverse velocities, and thereby, amongst others, energy spectra of transverse velocity increments. Whereas the measured velocity PDF’s are in full compliance with theory, the measured energy spectra look dissimilar. Several effects that contribute to this effect have been uncovered. The photon noise in the images adds a significant background to the spectrum over the full spectral range. Diffusion of the line mainly affects the inertial subrange, so that no k-5/3 scaling can be found. If we consider higher order statistics in the form of structure functions, we obtain much better quality statistics. Although the smallest scales remain affected by artifacts and diffusion, it does not significantly influence the larger scales. We find that for a write-read delay of 10 µs the scaling exponents of our transverse structure function actually correspond very nicely with the Log-Poisson model. An important characteristic of turbulence is its ability to transport and mix fluid effectively and this phenomenon can be expressed in terms of the evolution of the line separating two different marked regions of the flow. The tracking of such lines is nearly impossible by means of conventional techniques, but APART is well suited for this application. It is expected from theory that the line length increases exponentially with an exponent ??¿/t¿, where ??¿ is Re independent, but time dependent, as can understood by the model of Girimaji and Pope [77]. It has also been observed that the wrinkled line has a fractal dimension df > 1 and thus the stretching rate ??¿ found depends on the fractal dimension of the line and the length of the line elements ("rulers") that are used to determine the total length. The fractal dimension that is found is df = 1.017, a value that is comparable to our kinematic simulations (df = 1.020) but considerably lower than the value (indirectly) derived by Villermaux et al. [85] of df = 1.10. The experimentally found stretching rate, extrapolated to ruler size ¿, is found to have a maximum ??¿0 = 0.085, lower than those found in DNS simulations by Goto and Kida [89] (??¿0 = 0.17).