Detonation occurs because there's too much fuel with too little octane rating left after the charge is ignited. The purpose of adding extra fuel during high power operations is to prevent this by "overcharging" the fuel/air mixture and ensuring that the mixture is too rich to detonate and thus only burns. However, you can exceed the beneficial limit of this and get detonation anyway. The way to resolve it is to lower the amount of fuel in the charge to limit or eliminate the excess fuel in the after-charge, thus preventing the detonation by going lean.
However, by leaning, the cylinder temperature will increase because you are now burning more efficiently and completely with no fuel to evaporate after the main ignition.
Here's an article that I found on Detonation and some of the misunderstandings about it - http://www.avweb.com/news/pelican/182132-1.html
That link (and other articles regarding engine management under /news/pelican/) is interesting.
This is also interesting (from another topic):
I just found some real interesting test data on the Merlin XX, close enough for our purposes, and it indicates an air fuel ratio at or close to 13.7:1 (7% fuel air ratio) from manifold pressures as high as 50" Hg (20 psi boost) down to 20" (-5 psi boost) in rich mixture and low gear. That's actually leaner than the 12.5:1 AFR (8% FAR) best power setting in our diagram. That shows that the Merlin never actually enters a rich rich condition, ever, during their testing. No lean mixture data is available.http://www.enginehistory.org/members/ar ... sR-R.shtml
(table is at bottom of page)
But... what is the most interesting of all, is when you combine the graph from Pelican about mixture vs detonation tendencyhttp://www.avweb.com/newspics/pp43_deto ... ale_lg.gif
to the data on the Merlin XX (assuming early Merlins run as similarish air-fuel ratios)http://www.enginehistory.org/members/im ... ble-04.jpg
...you come up with the conclusion that running Merlin in "rich" setting gives you air-fuel mixture that's the most likely to detonate of all possible mixtures the engine could potentially receive (i.e just slightly richer than stoichiometric). Leaning out during high performance flight would indeed reduce likelihood of detonation, against common conception. This might be because running lean slows down the burning and reduces peak pressure after spark.
So, indeed, CAPFlyer might be onto something here. It would mean that both I and Scott were wrong... but modeling on A2A Spit with Core 1.4 would on the other hand be accurate because it clearly behaves exactly the opposite way as I and Scott would have assumed it to behave.