Modeling of mixing in cross junction using computational fluid dynamics

Research has shown that mixing in cross-junctions in water distribution systems is far from perfect, and that the entering fluids bifurcate from each other rather than mix. The purpose of this thesis is to study the behaviour of two fluids entering a cross-junction in a water distribution system. In this context, experimental tests and numerical simulations are performed in order to produce and test the mixing at cross-junctions. This study focuses on cross-junctions with equal pipe diameters, with flows that can vary from laminar to turbulent. The fluids are pure water and tracer. Different tracer materials with various flow configurations were tested experimentally and numerically. Firstly, an experimental study of mixing in cross-junctions was performed at the TZW: DVGW-Technologiezentrum Wasser (German Water Center) in Dresden. This experimental study pro-vides an overview of the parameters that can affect the mixing in cross-junctions, and is used to validate the numerical simulations. Different numerical approaches for modelling the mixing in cross-junctions are presented. The simulations use an existing commercial CFD code, ANSYS CFX 19.1, and are also extensively validated using experimental and numerical results from other researchers. In ANSYS CFX there are several models that can be used to simulate the mixing of two fluids. In this study both fluids are considered to be isothermal incompressible and without phase change. Two mixing models are tested: the additional variable model and the multi-component model. The three-dimensional models use RANS turbulence models and LES simulations. The parameters of the numerical setup were investigated carefully in order to study their effect on the results. Furthermore, the effect of changing the turbulent Schmidt number in the RANS simulations was extensively studied, and the results are compared with the experimental results. The accuracy of using Large eddy simulation to simulate mixing in cross junction is also tested, taking into consideration the required mesh resolution and the turbulence in the initial bound-ary conditions. This work presents an applicable numerical approach to simulate the fluid behaviours in cross-junctions. Using this approach, the effect of different parameters is tested, such as: Reynolds number, pipe diameter, mixing time, diffusivity and density difference. The results produced using the numerical approach revealed that one of the main parameters that affect the mixing is the density difference. It has a great effect on the outgoing concentration in cross-junctions, and the mixing behaviour changes when the tracer material and the flow regime are changed. The used approach will help to investigate the effect of various flow parameters on the mixing in cross-junctions. Based on the data set of this study, an empirical conceptual model for mixing in cross-junction is also presented using multiple regression, and there is potential for this model to be further developed in combination with experimental and numerical studies.