Meme of the week – only for CFD people ….

At Simulation of reactive Thermo-Fluid Systems (STFS), TU Darmstadt, we have a great little tradition: current events in the group get captured in our Meme of the Week.

Let's talk about those memes on grid size and what fits into RAM…

We all know that one classic reviewer question:

„Did you check grid convergence?“

or

„Is the grid fine enough?“

Yes – of course, we checked that. Initially.

But here’s some nuance that’s often lost in the punchlines:

Side note 1 – LES is tricky

In my graduate course Turbulence Modelling, I just discussed why grid convergence is not trivial in Large Eddy Simulation (LES).

There are rigorous methods (e.g., Richardson extrapolation), but metrics like „resolved kinetic energy > 80%“ can be misleading.

Numerous papers show: there is no universal answer. LES grid convergence is subtle.

Side note 2 – Higher resolution does not mean better statistics

Yes, high-res meshes produce beautiful snapshots.

But: do the statistics (what actually matters in LES) really improve?

Often, people push resolution limits without revisiting that question.

Also worth remembering: The subfilter model (e.g. eddy viscosity) and even the numerical scheme can significantly affect results — even on the same grid.

We published a study on this: same mesh, same numerics – just varying the model. The visual resolution looked very different.

https://lnkd.in/ewUj-WNp

So next time you see a “grid meme”, remember:

It’s not just about mesh size – it’s about what you resolve, how you resolve it, and what you conclude from it.