This is a Ph.D. dissertation. The aim of the present research was to explore and shift the limits of both numerical and experimental techniques for studying the coupling of chemical kinetics and hydrodynamics of turbulent reactive flows in industrial geometries. For this purpose, the turbulent liquid-liquid mixing of a reactive process stream in a tubular reactor (Re=4,000) with a perpendicular injector was investigated by means of three-dimensional numerical simulations and an experimental technique to measure the three-dimensional concentration field of mixing scalars. In the numerical approach, the large-scale turbulent motions in the tubular reactor were solved explicitly by means of a large eddy simulation (LES), while the transport and evolution of the reacting scalars were solved by using the transport equation of a filtered version of the scalar joint probability density function (PDF), denoted as filtered density function (FDF).

In this way, the advantages of both the LES and the PDF methodology were combined: the large scales are solved explicitly with models taking care of the influence of the non-resolved subgrid scales, without having to model the reaction rate term of the scalar components due to the closed form of this term in the FDF formulation. Due to the high computational costs involved, however, FDF/LES until so far has been applied to relatively simple flows only. In the present dissertation, it is shown that the current level of processor speed and memory storage of relatively cheap parallel Linux clusters has opened the feasibility of FDF/LES to tackle turbulent reactive mixing in process equipment. With FDF/LES, the mixing of the parallel competitive reaction components in a tubular reactor is studied. The yield of the slow reaction product as predicted from the simulations depends in a physically consistent way on both the Damkohler number and the inlet concentration of the feed-str! eam.