Computational fluid dynamics (CFD)

Computational fluid dynamics is the numerical simulation of fluid flow, air, water, or other gases and liquids, governed by the Navier-Stokes equations, used to predict how velocity, pressure, and temperature vary around or within an object.

The fluid domain is divided into a mesh, and the governing equations are discretized and solved iteratively across it, an approach similar in spirit to finite element analysis but centered on flow rather than solid deformation, and typically needing much finer resolution near surfaces to capture turbulence accurately. Weather and climate models rely on closely related numerical methods, applied at a very different scale.

CFD is used across aerospace, automotive, and HVAC design, and increasingly as part of a digital twin of a physical system, running on high-performance computing clusters because realistic simulations are computationally demanding. Pitfalls center on turbulence, which cannot be solved exactly at full resolution and must instead be approximated with a turbulence model; the choice of model and the quality of the mesh both strongly affect accuracy, so CFD results are normally validated against wind-tunnel or field measurements rather than trusted on their own.