The whole LiPo fireball myth has been elevated to almost cult status, now, it seems.
I spent some time trying to track down and separate genuine lithium battery fire cases from mythical ones to see just how probable it was that these cells would catch fire in normal use. The first thing I discovered was that the number of incidents is very small, just a few dozen worldwide. As far as I can tell, the total number of lithium battery sales runs into the hundreds of millions.
All of the incidents of cell problems on discharge that I could find related to very early lithium cobalt oxide cells with a low maximum discharge rate, being discharged at close to, or in excess of, the manufacturers stated maximum. None resulted in actual fires or explosions, but there seem to have been three incidents where cells got hot enough to vent and release fumes and vapour. Since the development of high discharge rate cells the discharge related incidents seem to have become a thing of the past. The worst that happens is that the cells die and may puff up a bit.
There are reported instances of fires caused by batteries being short circuited by physical damage, but these relate to all battery chemistries. Most battery explosions seem to occur with lead acid cells, most battery fires seem to occur with nickel and lithium chemistries. I think it's fair to say that any battery presents a big risk of fire or explosion if shorted out.
The vast majority of lithium cell incidents occur whilst charging. It's hard to be definitive, but it seems likely that defective chargers or charging at too high a rate has been the major cause of incidents. Since the near-universal adoption of chargers with a proper CC/CV charge characteristic these incidents have reduced dramatically.
The user group that place the greatest demands on LiPo type cells are the RC model users. They frequently discharge at rates of around 5 to 6C average and expect to charge at rates of at least 2 to 3C in the field. They rarely use cell-level discharge monitoring, instead relying on the very crude cut-off voltage that can be programmed in to most RC typoe controllers via the beep codes or programming cards. Many users leave their controller at the default setting (which is often a very low LVC) and don't bother with programming them, because it can be a pain to do on the cheap ones. The result is that incidents like Bob's are not that uncommon, yet they rarely cause any risk of damage, other than killing the battery pack.
A scan around the RC forums reveals that battery fires are often discussed but rarely witnessed or factually reported, other than people deliberately trying to set packs on fire, a bit like Bob's experiment.
All the evidence points to LiPo packs that use lithium cobalt oxide being reasonably safe and very unlikely to catch fire or explode if properly managed. Careful charge control is vital, and proper cell-level discharge monitoring prevents early cell failure. Cell level monitors are very cheap and effective, something like the $2 cheap monitor from HK (
http://www.hobbycity.com/hobbyking/stor ... oduct=4175 ) or the more capable $12 one (
http://www.hobbycity.com/hobbyking/stor ... oduct=8927) would have probably saved Bob's pack from cell failure.
Jeremy