COMSOL 6.4 - Water Purification Reactor

Water purification for turning natural water into drinking water is a process constituted of several steps. At least one step must be a disinfectant step. One way to achieve efficient disinfection in an environmentally friendly way is to use ozone. A typical ozone purification reactor is about 40 m long and resembles a maze with partial walls or baffles that divide the space into room-sized compartments (Ref. 1). When water flows through the reactor turbulent flow is created along its winding path around the baffles toward the exit pipe. The turbulence mixes the water with ozone gas that enters through diffusers just long enough to deactivate micropollutants. When the water leaves the reactor, the remaining purification steps filter off or otherwise remove the reacted pollutants.

When analyzing an ozone purification reactor, the first step is to get an overview of the turbulent flow field. The results from the turbulent-flow simulation can then be used for further analyses of residence time and chemical species transport and reactions by adding more physics to the model. The current model solves for turbulent flow in a water treatment reactor using the Turbulent Flow, k-ε interface.

Model Definition

Model Geometry

The model geometry along with some boundary conditions are shown in Figure 1. The full reactor has a symmetry plane which is utilized to reduce the size of the model.

Figure 1: Model geometry. All boundaries except the inlet, outlet, and symmetry plane are walls.

Domain Equations and Boundary Conditions

Based on the inlet velocity and diameter, which in this case correspond to 0.1 m/s and 0.4 m respectively, the Reynolds number is

Here ν is the kinematic viscosity. The high Reynolds number clearly indicates that the flow is turbulent. This means that the flow must be modeled using a turbulence model. In this case, the k-ε turbulence model is used, as it is often done in industrial applications, much because it is both relatively robust and computationally inexpensive compared to more advanced turbulence models. One major reason to why the k-ε model is inexpensive is that it makes use of wall functions to describe the flow close to walls instead of resolving the very steep gradients there. All boundaries in Figure 1, except the inlet, the outlet, and the symmetry plane, are walls.

The fully developed flow is used as inlet boundary condition. A constant pressure is prescribed on the outlet.

Notes About the COMSOL Implementation

Three-dimensional turbulent flows can take a rather long time to solve, even when using a turbulence models with wall functions. To make this tutorial feasible, the mesh is deliberately selected to be relatively coarse and the results are therefore not mesh independent. In any model, the effect or refining the mesh should be investigated in order to ensure that the model is well resolved.

Results and Discussion

Figure 2 shows the velocity field in the symmetry plane. The jet from the inlet hits the top of the first baffle which splits the jet. One half creates a strong recirculation zone in the first “chamber”. The other half continues down into the reactor and gradually spreads out. The velocity magnitude decreases as more fluid is entrained into the jet.

Figure 2: Velocity field in the symmetry plane.

Figure 3 gives a more complete picture of the mixing process in the reactor. The streamlines are colored by the velocity magnitude, and their width are proportional to the turbulent viscosity. Wide lines hence indicate high degree of mixing. The turbulence in this example is mainly produced in the shear layers between the central jet and the recirculation zones. The mixing can be seen to be relatively weak in the beginning of the reactor. The plot also shows that it increases further downstream.