Duvar yakınındaki türbülans için yalın bir formül

Oxford engineers have combined machine learning with fluid physics to produce a compact formula for a difficult turbulence problem. When air or water moves along a surface, turbulent motion gains energy from the main flow while losing energy as heat. These rates can differ sharply near a wall, yet many models treat them as equal. The researchers used symbolic regression, which searches data for mathematical expressions, to describe how the ratio between production and dissipation changes with distance from a smooth surface. Their equation has only four constants. After fitting it to channel-flow simulations, they tested the same constants without refitting against 19 benchmark profiles covering channels, pipes and boundary layers. It closely followed the tested near-wall data and reduced error in estimating dissipation by 88–95 percent compared with the equal-rate assumption. The formula and code offer researchers an inspectable, reusable reference. However, the study does not yet demonstrate that the method improves complete engineering flow simulations.