I don’t generally watch much sport – I like F1 more than most things, but since Wimbledon is just 5 miles down the road from me its hard not get swept up with tennis fever which grips the nation for two weeks every year and the hope that Andy Murray will make it to the final. Sadly that was not to be, but the way Andy Roddick played in the final yesterday was just phenomenal. He was unlucky not to win.
In the final set, Federer returned a deep lob from Roddick with an overhead shot that looked effortless. And wayward. But it was hit with enough side spin to bring it back well inside the sideline – maybe as much as 30cm, whereas from the angle I saw it hit, it looked like it would go out by at least as much. The cloth covering, and the deformation of both the ball and the racquet strings make it possible for the top players to get the ball to spin at over 3000 rpm, but what fascinates me is how much the spin affects the ball’s trajectory.
The spin affects the boundary layers that form on the surface of the ball, the location of the stagnation point and the wake behind the ball by affecting where the boundary layer separates from the surface. Unlike a football or a golf ball, a tennis ball is not smooth. The surface is very rough, increasing the effect of the spin on the surrounding air flow.
In conventional boundary layer theory, the roughness of the surface only affects the flow if the roughness disturbs the laminar sub-layer – the thin layer of fluid that moves smoothly over the surface, even if the flow is turbulent. For this to happen, the roughness has to be approximately the same size as the sub-layer thickness.
Roughness also plays a role in electronics cooling, increasing both system pressure drop and heat transfer. Most surfaces in an electronics system can be considered rough when viewed in the context of the boundary layer concerned. On a PCB, flat pack components protrude up into the air flow, disturbing the boundary layer that forms from the leading edge of the PCB. Boundary layers also form on the tops of these packages, and the plastic encapsulant can also be rough affecting heat transfer at the leading edge of the package where the boundary layer is very thin.
Of course this discussion about boundary layers ignores how complex the flow over a PCB really is, but hey that’s what we have CFD for. What’s important is to describe the situation correctly by defining how rough the various surfaces are. Like all software, CFD tools work on the GIRO system – Garbage In, Rubbish Out.