Mesh Adaption in Ansys Fluent - Part 1

Mesh Adaptation in Ansys Fluent - Part 1 Hello everyone, this is Mohsen Seraj, the senior application engineer at Ozen Engineering. Today, I want to talk about mesh adaption, specifically geometry-based mesh adaption in Ansys Fluent. We have two videos on this topic.

In this first video, I will discuss the model setup for mesh adaption, compare the results with manual refinement, and show the workflow for geometry mesh adaption. In the workbench project, we start with a simple geometry, a quarter of a pipe with sharp points.

We have walls, inlet, and outlet faces. We set up the size for periodic boundary conditions and add name selections for staples. We will create a pattern ruler for the mesh and set the axis and side set for the oval almond shape. For the mesh setup, we will use Fluent Meshing.

We consider velocity inlet, pressure outlet, and periodic boundary conditions. We set up the mesh for the minimum and maximum sizes for the periodic conditions. We describe the geometry boundaries and consider velocity inlet, pressure outlet, and periodic boundary conditions.

We check the mesh quality for orthogonality, which is quite low. However, the initial mesh has a good quality based on the minimum orthogonality of 0. 255. We have approximately 2586 cells. We run the model again and check the mesh quality.

The mesh quality is 0.255, which is above the minimum acceptable orthogonality of 0. 15. We have 2586 cells. We look at the edges for the inlet and outlet. We have six segments for the curved part of the domain. The boundary layer has only three cells, which is not adequate.

We define the material as water from the Fluent database library. We set the constant properties for density and viscosity. We set the boundary conditions for inlet, outlet, and periodic boundary conditions. We define the expression for calculating the pressure drop between inlet and outlet.

We set up the solution for 500 iterations and run the model. We check the residuals, cell count, and pressure drop. We move on to the refined mesh, which has approximately 6000 cells. The minimum orthogonality is 0.27, which is good. We look at the edges for the inlet and outlet.

We have more cells at the sharp corner, which is good. We initialize the model and run it for 500 iterations. We check the cell count and pressure drop. The pressure drop is approximately 41,000, which is a 10% difference from the initial mesh. We move on to mesh adaption.

We have done the initial mesh and manual meshing, which took approximately 2500% of the time. The pressure drop is above 45, 000. The refined mesh has approximately 6000 cells, and the pressure drop decreased to 21, 000. We work on the original mesh and see how mesh adaption can help.

We start setting up the mesh adaption based on two criteria: one for having more information layers into the boundary layer at the walls, and the other for adapting the bulk of the fellow based on the pressure gradient.

We define the criteria for refinement for the boundary layer and use the wall boundary zones. We choose at least two cells for the boundary layer. We save and display the cells in the register. We have approximately 1000 cells out of 2500 cells.

We set up the mesh adaption for the bulk of the fellow based on the gradient of the pressure. We name the cell register "pressure" and set the range for the maximum and minimum of the pressure gradient. We run the solution for some iterations and compute the pressure gradient.

We set up the register for the pressure gradient and say that the minimum is 100. We see that the cells are marked for the pressure gradient. We set up the automatic mesh adaption based on the boundaries and pressure gradient. We set the level of refinement to the minimum of 500,000 cells.

We set the minimum orthogonality after mesh adaption to the default values. We set up the solution reports for cell count and pressure drop. We create a plane for the mesh refinement at the shock. We start the mesh adaption simulation for 500 iterations.

The cell count jumps to approximately 200,000 cells, and the pressure drop is approximately 44, 450. The mesh quality is low, and the mesh adaptation did not improve the pressure drop calculation. We need to have a good initial mesh that can capture the physics of the problem.

We need to have enough cells to capture the geometric features and good representation for the fellow features. In the second part of this video, we will discuss geometry-based mesh adaption and show how we can improve the results. Thank you for watching.