Once I had a Cherokee in a group of 10 that was stored in an "open" hangar during the month of December when it rained the whole month.  The previous user left the tanks empty.  Apparently the tanks collected a lot of condensation (about a pint it turned out).  When I took it the temp was well below freezing and I got fuel from a truck that had been parked outside.  There was no trace of water in the fuel.  I flew it to Moline, which was above freezing and I filled it from an underground tank.  Still no significant water - I think the water hadn't run to the bottom yet.  Engine lost power in the soup at altitude over Iowa City.  My conclusion was that the vibration from the engine gradually jiggled the water droplets to the bottom, when they went into the gascolator, which happened to be in a box cooled by outside air.  The water froze on the filter screen shutting off the fuel flow.  The electric pump, being in series with the filter, couldn't help.  Later, I asked an engineer at Piper why they would put the electric pump, which had an integral sump and filter, in series instead of in parallel with the gascolator.  The answer was they felt that the mechanic might not drain the sump and check the filter during the annual.  Little consolation to me.  Bottom line was that the gascolator hung on to much of the water, possibly preventing a much earlier and more complete power loss.  I'll keep my (oversize, by the way) gascolator, thank you.  And now I rock the wings before checking for water, especially in a Cessna.

Regarding filters, Bendix had trouble deciding whether to put a bypass around the inline filter in their system.  At first they put a bypass valve in it (actually the filter pushed itself off the seal against a spring.  They got sued because someone kept going with a plugged filter until the crud plugged the nozzles.  So they turned the filter around so that it wouldn't bypass.  They got sued because the filter plugged and there was no emergency bypass.  I think they all have a bypass capability now.  I would NOT put any extra hardware upstream of the engine-driven pump.  One could argue that a plugging filter would just result in a power reduction, not a stopped engine.  However, the more likely failure mode might be a pressure drop across the filter that would cause the engine pump to vapor lock, resulting in a complete engine failure.  Sure, there is an electric pump, but only if the pilot turns it on quickly enough.  And why add a filter that is finer than the downstream orifices?  I'll stick to the standard hardware.