Free surface turbulence: Coherent structures and mass transfer estimation

This thesis investigates the nature of turbulence at free surfaces in open channel flows. The experiments are based on Ludwig Prandtl's observations, in 1903, of vortices (coherent structures) at the free surface in open channel flows. The aim here is to discover the various forms that turbulence takes at shear free interfaces. Streak photography, Digital Particle Image Velocimetry (DPIV) and Direct Numerical Simulation (DNS) were used to classify the free surface structures, which were found to be vortices, upwellings, and downdrafts. Streak photographs revealed the dynamics of free surface turbulence, which involve interactions among the various structures, and indicated the persistence of the vortices. It was found that these structures follow a mixed scaling with regard to the inner and outer variables of the underlying flow field. DPIV provided quantitative measures of the various features of free surface turbulence. Following McCready et al. (1986), an estimation of the scalar transfer coefficient was found to over predict by a factor of 2, the CO{dollar}sb2{dollar} mass transfer measurements of Komori et al. (1989). DNS provided a simulation of the fully non-linear mass transfer equation. Two approaches were considered. The first approach involved solving the non-linear mass transfer equations near the interface using numerical data. In the second approach, however, the velocity fields generated by DPIV were input into the DNS, and the simulation was run for the mass transfer near the interface. In both approaches the mass transfer coefficient was estimated and compared with the experimental measurements of Komori et al. (1989). For Reynolds number 2800 and depth 1.5 cm, the mass transfer coefficient in the purely numerical case was found to be 2.81867 e-5 m/s, and in the case with DPIV velocity fields to be 2.06687 e-5 m/s. This is comparable to Komori et al's experimental value of 1.65 e-5 m/s. For Reynolds number 2800 and depth 2.5 cm (DPIV velocity fields), the mass transfer coefficient was estimated to be 1.19413 e-5 m/s, which is again comparable to Komori et al's experimental value of 0.75 e-5 m/s.