Well, knocking or dropping a heated case in water makes steam as it quenches, but no explosion. The water is a phase change material, same as the lanolin in Heat Stop or the other plumbing heat blocking pastes, so it will carry off heat as steam if convection doesn't replace the hot water with cold fast enough. You'll see little bubbles nucleate on the surface of the case same as you see in the bottom of a pan getting near a boil, and they'll rise and go off as steam. Heating liquid water one degree only takes one BTU per pound, but the
enthalpy of vaporization of water is over 970 BTU/lb, so turning it from 212°F liquid to 212°F steam absorbs a lot more heat.
Heat transfer can be a confusing topic. Heat does move toward cold, but only because the temperature difference between hot an cold exists. It is driven by the difference in temperature, somewhat analogous to electric current being driven by a difference in voltage, which is why thermal resistance and conductivity are standard thermal properties with standard units and why temperature drops can be calculated in thermal resistances exactly by the same rules as voltage drops in electrical circuits are. There are differences, though. One of them is that the flow of electricity through an electrical resistance dissipates energy which makes heat. Heat flow through a thermal resistance doesn't make anything new. It does change the infra red color emitted as it changes temperature by flowing in or out of an area, but otherwise it is just moving.
So, there is no reason to object to moving heat through a case toward the head. It was going to go there by thermal effusivity if you had no chill on the case head anyway. The difference is that with the cold object there, the heat will continue to flow because the cold object keeps the head colder than the heated neck. If you remove the cold object, heat flow decreases, but only because the head is getting warmer, thus reducing the temperature difference that drives the heat flow. That's not a desirable condition. It is actually better for the head to stay cold enough to cause constant flow than to let it heat and slow the flow.
Water up near the shoulder will cause trouble, too. It will keep the case at that point at about the boiling point of water with all that energy the conversion to steam absorbs. It will do that faster than the torch can put heat in through the thin brass. As the steam goes off, unboiled water flows in to displace it due to its weight under the force of gravity. That will cause the brass to remain too cold to stress relieve for some distance above the water line due to the temperature drop.
Here's another point. You don't actually need to fully anneal brass. As mentioned earlier, that makes it too soft due to grain growth. All you need to do is relieve the stress from work-hardening that causes splits. That is complete right at the bottom end of the annealing temperature range. I put a chart below showing how hot brass has to be to stress relieve and anneal in one hour. Unfortunately its legend doesn't specify how thick the samples are that it was derived from, so you can't tell how much of that hour is actually needed to produce the plotted effects, and how much is just the soak time needed to bring the samples up to temperature through and through. Mostly we stress relieve cases by making them hotter than the stress relieving range in the chart so they will stress relieve faster, but not so hot that portions of the mouth get over soft. It's a compromise.
Stress relieving occurs between 482°F and 572°F on the chart. 662°F is where the softening grain growth starts. 650°F seems to be the generally targeted range for case annealing. We want good neck tension for good start pressure, as that just avoids the grain growth (see grain growth plot in the chart below). Once the brass is hot enough to start grain growth, it will continue until the temperature drops, which is why quenching in water will stop grain growth. However, brass cases cool fairly quickly in air, so I doubt failure to quench makes a lot of difference to them. If you stay below 662°F it won't matter because you aren't initiating grain growth anyway. You can set cases down on an aluminum plate with a fan blowing over them if you want to speed cooling without quenching.
I still quench, but maybe I'm just being superstitious? I'm using a 650° Tempilstick on the necks and I've not done any testing to find any difference the quenching makes. I'm sure that by the time the indicator wax has melted some portion of the mouth has exceeded 650°F and started grain growth—but significant difference?
It would not be very hard to set up to try to measure it. Bullet seating and pulling force should indicate whether the grip on the bullet is greater for one than the other, indicating softer or harder brass to a degree that will affect shooting performance, and I have instrumentation for measuring that. I"ll try to remember to test that next time I have some inexpensive brass to stress relieve so I can sacrifice a few cases. Seating and pulling force before and after the heat treatment should quantify the effects of the different methods.
