DEMO/ INSIDE THE PLATFORM

Look inside the workspace.

One student’s project, four rooms. Click around: open a lesson, move the refinement slider, tick a milestone, join the thread.

Lesson 4 of 8 · Computational fluid dynamics

Adaptive mesh refinement principles

How this unit was checked 16 of 16 code examples ran · technical review corrected 1

Core theory

A uniform mesh wastes cells where the flow is calm and starves the wake where it isn’t. Adaptive refinement splits only the cells where the solution changes fastest. For incompressible flow we solve

(1)∇ · u = 0
(2)∂u∂t + (u · ∇)u = −1ρ∇p + ν∇2u

On each cell K of size hK we estimate the local error with a vorticity indicator, then flag the worst cells:

(3)ηK = hK ‖∇ω‖K,   refine K if ηK > θ · maxJ ηJ
0 cells · 0 refined

Halving h on a scheme of order p cuts the error by about 2p, which is why refining the right fraction of cells can beat refining all of them:

(4)eh ≈ C hp  ⇒  eheh/2 ≈ 2p

A preview with sample data. Names, posts and applications are illustrative; nothing you type here is sent anywhere.

Want a workspace of your own?

Register interest Questions first? Read the FAQ →