The print failures are evenly spaced so I would assume that whatever is causing them occurs in a cycle. The 5mm part does not seem to have any problems which makes it harder to diagnose. Maybe the 5mm thickness creates enough of a support barrier to prevent sagging. Or the print time for those layers is longer, which allows the PLA to cool before the next layer is applied. I can't pinpoint anything so I toss out these crazy ideas.
1. Your reel is binding at some point in the rotation. The extruder doesn't get enough filament as it struggles to pull the reel free but once it does, the reel rotates around easily until it cycles to the binding spot again.
2. Your filament is damaged in some way at one spot in the reel. Every time the reel gets to that part in the loop, the problem occurs then goes away.
3. Your extruder clamping gear is stripped or packed with filament grind in one spot. When it rotates around to that spot the filament slips a bit resulting in a lower federate. (This is unlikely because the gear is small so the spacing of the problem would be much closer.)
I can't think of why the fan would cause this problem, unless it is turning off at evenly spaced time intervals. I suppose if you had an intermitted short in the fan wiring, it could heat up, break the circuit, cause the filament to overheat and sag, cool off, reconnect the circuit and cool the filament properly until the fan heated up again.
I've got an E3d also and need to update my fan shroud. If you happen to design a new shroud consider posting it.
Chris