Expansion fans on exhaust valve seats.
(Fluiddynamics)During the blowdown phase, as the exhaust valves open, the gas entering the exhaust ports is just about supersonic, especially at low lift angles. This supersonic flow is susceptible to expansion fans.
https://en.wikipedia.org/wiki/Prandtl%E2%80%93Meyer_expansion_fanThis is why radiused exhaust valve seats and tulip valves with a rounded valve edge on the CC side helps flow. Expansion fans accelerate flow and are isentropic, that means without generating entropy in the way that a normal or oblique shockwave does. Not only does this simplify the calculations, but the expansion fans have very little losses. Typically a shockwave acts as a barrier for air, and causes non-reversible losses. An expansion fan on the other hand eliminates MOST, not ALL losses up to the accelerated velocity of ~Mach 1.4.
The angle needed to achieve M1.4 via expansion fans is ~9-10 degree steps. The air that flows through the seat and the valve face is flowing through a series of angles, if these angles can be kept to 9-10 degrees, one can dramatically lower losses on exhaust blowdown.
This is a lot harder to achieve on the intake valve(s) because the flow there never goes above Mach .55 maybe locally at low lifts with high VE it could get close to M=.6-.7 or the trans-sonic regime. Although for port injection, one also has to consider wet flow(ie fuel) and as a result limits the angles that can be used depending on the engine application. So while a full radius job on the intake will flow more CFM, it won't necessarily make more power, because smooth angles not only allow air to stick, but also any other fluid, such as fuel.
https://i150.photobucket.com/albums/s92/chippievw/F1technical/F1technicalintakeport002_zps9d11215f.jpgHere you can see a cross section on the valve seats of a Cosworth V-10 F1 engine. Exhaust on the right, intake on the left. You can see the curve on the exhaust port is more pronounced compared to the relatively straight intake port, which only has a slight curve near the valve seat. This turn is necessary to guide airflow across the seat and into the combustion chamber, as well as to allow fuel and some air to separate from the wall to avoid pooling. Of course, at 18,000rpm the air is already mostly turbulent and will mix the crap out of the fuel. However at part throttle and lower rpm, the flow can be so laminar that fuel sticks to the walls and falls out of suspension.
For this reason engines that operate at lower RPMs and are undervalved(ie not enough valve for the combustion chamber), will benefit from 5 discreet angles on the intake side, particularly if the engine is carbureted.
https://www.speednik.com/wp-content/blogs.dir/1/files/2013/05/Valves_2.jpgWhile such a setup will flow a bit less than a full radiused valve seat, they typically work better in terms of combustion stability, and can help to limit flow reversion. On the exhaust seats, there is usually no fuel being burned(unless you're doing anti-lag), and you want the flow to leave the CC as fast as possible, doing full radius helps. You don't have much reversion on the exhaust seats, because the exhaust manifold is at a much lower pressure, and the overlap helps maintain the pressure gradient towards the exhaust.
With direct injection, since you don't have to worry about fuel in the intake, the rules change somewhat, and what is beneficial for the exhaust side, is also beneficial for the intake side as fuel is injected directly into the CC. When boost is introduced the similarities and ability to exploit expansion fans starts to work on the intake side because air density has a small but measurable effect on flow velocity. This can leave the air smack dab in the transonic region especially at low lift levels. Since valves open and close, the valves are at low lifts twice, and high lift once.
