I'm not an engineer, a chemist, or even an experienced modder, but I do know a couple things about systems, and how things fail.
There are plenty of 'every time' failures - when you take one of stock part 'A' and one of 'B', and go through a particular process, you get a failure. Even if it takes a long time and a bunch of work to determine what 'A' and 'B' (or however many batteries at various charge levels, flashlights, or whatever) are, and what the steps or conditions are, you get a failure more-or-less consistently once you've got it worked out.
These are hard to find, but once you've got it hammered out, it's reasonably straightforward to describe what not to do.
Then there are the ones that turn out to be based on variability in the parts involved - in this case, it could be that it takes (for example - I don't know enough to promote a hypothesis), mismatched batteries, one of which has high water content AND a slightly sticky vent so that air expansion during air transportation has squeezed the electrolyte around a bit, the other of which has a high internal resistance, in a light with a stiff spring that was made on the suppliers third machine from the left, which has always cut a little longer than the others.
This type of 'cascading' failure, in which no one item is the 'cause', but when they all line up, you get an unusually high rate of failure, seems the most likely to me for this situation. It may be that the PM6 has an unusually stiff spring, or that some brands of batteries (or ones made on Mondays, or whatever) have more variability, but it's only when they all come together that you get a significant 'event'. It makes testing very very hard, and takes using a fairly large number of samples to get enough 'mix and match' to see what's going on.
Then comes trying to figure out whether you've got a truly 'random' failure - in which battery number 999,685 just came out of the machine wrong, and just happened to match up with flashlight number A7843, which was threaded a touch longer, or if there's some kind of systematic failure, in which significant numbers of lights or batteries share the characteristic that contributes to failure.
This is the kind of question that drives Six-Sigma consultants into a seizures of glee, as they calculate the fees they will charge to assist you in determining the type and magnitude of your variability, and imagine themselves flying back and forth to China in 1st class on your dime a half dozen times before dropping an inch-thick report that explains that you DO have excessive variability, and you should attempt to improve, which you already knew in the first place. They'll be happy to help you with implementing that project, for only a wee retainer...
Kudos to both the OP, both companies involved, and everyone working to assist - it's a joy to watch folks working on solving a problem together, instead of poking each other with sticks and shouting insults.
There are plenty of 'every time' failures - when you take one of stock part 'A' and one of 'B', and go through a particular process, you get a failure. Even if it takes a long time and a bunch of work to determine what 'A' and 'B' (or however many batteries at various charge levels, flashlights, or whatever) are, and what the steps or conditions are, you get a failure more-or-less consistently once you've got it worked out.
These are hard to find, but once you've got it hammered out, it's reasonably straightforward to describe what not to do.
Then there are the ones that turn out to be based on variability in the parts involved - in this case, it could be that it takes (for example - I don't know enough to promote a hypothesis), mismatched batteries, one of which has high water content AND a slightly sticky vent so that air expansion during air transportation has squeezed the electrolyte around a bit, the other of which has a high internal resistance, in a light with a stiff spring that was made on the suppliers third machine from the left, which has always cut a little longer than the others.
This type of 'cascading' failure, in which no one item is the 'cause', but when they all line up, you get an unusually high rate of failure, seems the most likely to me for this situation. It may be that the PM6 has an unusually stiff spring, or that some brands of batteries (or ones made on Mondays, or whatever) have more variability, but it's only when they all come together that you get a significant 'event'. It makes testing very very hard, and takes using a fairly large number of samples to get enough 'mix and match' to see what's going on.
Then comes trying to figure out whether you've got a truly 'random' failure - in which battery number 999,685 just came out of the machine wrong, and just happened to match up with flashlight number A7843, which was threaded a touch longer, or if there's some kind of systematic failure, in which significant numbers of lights or batteries share the characteristic that contributes to failure.
This is the kind of question that drives Six-Sigma consultants into a seizures of glee, as they calculate the fees they will charge to assist you in determining the type and magnitude of your variability, and imagine themselves flying back and forth to China in 1st class on your dime a half dozen times before dropping an inch-thick report that explains that you DO have excessive variability, and you should attempt to improve, which you already knew in the first place. They'll be happy to help you with implementing that project, for only a wee retainer...
Kudos to both the OP, both companies involved, and everyone working to assist - it's a joy to watch folks working on solving a problem together, instead of poking each other with sticks and shouting insults.