[TJP]

Hello everyone!

I am new to CFD and I am trying to simulate airflow over a cylinder (2D). I am having difficulty in understanding how to do mesh convergence. The mesh I currently have is unstructured but around the cylinder, there is inflation where the mesh is structured. How would I go about refining the mesh to show that the accuracy of the result converges

[LuckyTran]

Step 1) Change the mesh
Step 2) rerun it
Step 3) compare the result
Step 4) repeat steps 1-3 for several meshes


Mesh convergence isn't a single step study. Systematic variations in the mesh are not so straightforward. You can change only the inflation layers, or leave the inflation layers alone and change the core mesh or both or neither

[TJP]

So, if I was to plot a graph showing the convergence, what would be on my x and y axis ? Thanks

[LuckyTran]

Originally Posted by TJP

So, if I was to plot a graph showing the convergence, what would be on my x and y axis ? Thanks

It wouldn't be a single x axis but like a 15-way plot of x1,x2,x3,..... xn. But let's say you pick 1 x value and that's your total cell count (because that's one parameter that is quite universal). The y-axis should be your sought after parameter. Normally people pick 1 value for y but just like x's, there can be many many y's that you care about.

[karachun]

Usually people check some integral values like Lift/Drag coefficient, torque on turbine/compressor, pressure drop, mass flow (if you have pressure inlet and outlet), average temperature at outlet, quantity of heat energy transmitted to fluid/solid or frequency of some transient features like von Karman vortex.