April | 2018

This project was given in my ME260B Advanced Fluid Mechanics class as a midterm project. Our goal was to select a flow pattern and recreate it using a CFD (Computational Fluid Dynamics) package of our choice. We needed to both recreate the flow as well as create the flow at a Re (Reynolds Number) of 100. Through this, I got my first experience with ANSYS Fluent.

First, I defined my geometry. This was relatively straightforward through the built-in ANSYS geometry tool. I then had to define a mesh to analyze. Given that my computer was powerful enough and the geometry simple, I let ANSYS lay a fine mesh, seen below.

In the next step, I needed to set my boundary conditions. I defined that along all walls, the walls were stationary and had a no-slip shear condition. At the inlet, I used a velocity magnitude normal to the boundary as my condition. This was where I could set the Reynolds number of my CFD. At the outlet, I merely set the backflow direction normal to the boundary using total pressure. I then set my experiment to compute from the inlet.

I assumed that the pipe was sufficiently long in each direction from the bend. The uniform velocity inlet would have sufficient distance to fully develop into the expected parabolic flow pattern before reaching the bend. In addition, I was only interested in the immediate area around the bend, namely the flow separation.

I defaulted to using air as my fluid, but did calculations to account for the differences in kinematic viscosity in maintaining the Reynolds number.

To visualize my results, I wanted to see both a velocity magnitude map as well as streamline patterns. Because the Re of 100 was very laminar flow, I was able to use a steady state solution. For Re = 2000 however, I needed to use a transient state solution to view the turbulence structures. The Re = 2000 test case was able to sufficiently recreate the notable structures of the original image.