When the seal between the valve and seat is broken, superheated combustion gasses escape between the valve and valve seat (pressed into the cylinder head casting) and the valve overheats rapidly. The escaping combustion gasses are as hot as a torch flame, and act literally like a oxy-acetylene cutting torch. If the leak last long enough, the valve starts to glow, overheat, and chunks blow off—just like a cutting torch flame does to a piece of steel. That is what happened to this exhaust valve:
The contact between the exhaust valve and seat is important for two reasons. The first is obvious: To seal so compression is achieved and combustion possible. The second is to keep the valve from burning up. The valve relies on the contact with the seat to keep cool. You are probably saying, “Yeah, right. How on earth can a valve moving so quickly spend enough time closed to cool off!?” This is possible because the valve spends more time closed than open.
Our water boxer cam profile, for example, has the exhaust valve open for around 220 out each 720 degrees of crankshaft rotation required to complete all four strokes. This means it is closed (720-200/720=) 69.4% of the time. And, that is the way it stays cool: when closed, heat flows from the very hot valve to relatively cool valve seat, which is liquid-cooled in this, and in most cases. If the valve cannot get rid of heat through the seat, it will burn up rapidly.
You would think this could not possibly work, at least not reliably—but it actually does. Burned valves on modern engines is practically unheard-of. Even on the not-so-modern water boxer design, it is also the case. On the over 3000 engines we have built, we have installed over 12,000 exhaust valves on the pair of heads that went on each engine. We have shipped out thousands more exhaust valves on the heads we have sold ala carte. In all those valves, we have had maybe a half dozen burned exhaust valves reported. Even if only one out of ten that burn were reported, and the failure was 10 times that (+/- 60), the failure rate is still miniscule. But, why does it happen, at all?
When you are looking at a burned valve that has been running for a while, it is the egg you see, not the chicken. The chicken that laid that egg comes in two forms: poor parts and/or workmanship or something keeping the valve from closing fully.
When the cause is poor parts or workmanship, it will not likely happen infrequently—it is usually a systemic problem within the supply chain. This was the case with AMC heads when they were first introduced. When parts or workmanship are the cause, the failures are not confined to just one valve every once and while—it happens a lot, and word gets around.
The more common reason a valve burns, and why it is so rare, is when a piece of carbon gets stuck between the valve and seat. This is typical of engines that are either running way too rich (too much fuel), or are burning too much oil. Either condition results in an excessive amount of carbon build-up up in the combustion chamber, which flakes off regularly. If the conditions are right, a peace will get wedged between the seat and valve just long enough for the blow-torch action to start a small nick on the valve where it meets the seat. Once a leak is introduced, it is just a matter of time before the blow torch affect takes over…
Chicken never tasted so bad!