What I don't get here is why changing the platen gap would solve the issue if it were due to a paper feed speed problem?
Platen gap could also compensate paper feeding for heavier substrates based on some Epson estimates.
Almost all wide format inkjets measure the paper feeding on the paper feeding axle, that method isn't 100% accurate as the paper weight differs, the friction differs per paper texture and paper softness, the roll can be wounded tight or less tight, paper rigidity influences the curl that has to be straightened and humidity can vary all the parameters mentioned first. So with known paper qualities compensations are added. A far nicer approach is measuring the actual paper speed with a sensor that checks the paper texture movement. The HP Z6100 has that but no other wide format that I know off.
Platen gap could also give a slight expansion of the droplet pattern, covering up the small mismatches between head strokes and paper feeding or similar flaws.
What I recall of the attached image (it doesn't appear anymore) didn't make me think it is a paper feeding issue. That usually translates to smaller bands/lines that are not covered with the right density. And I disagree with the opinion that that fault becomes progressively worse on longer lengths. If so then that is an indication of slip on the transport axle where the printed paper at the front wheighs in so to say. Your sample showed almost 1:1 banding so a bidirectional printing issue is more likely. That it starts at some point in the printrun is an indication that ink starvation is the main cause. That printing at a slower speed and/or unidirectional improves the print quality is another sign.
Either Epson is at the edge of the heads capacity or ink feeding capacities with your settings of speed and finest detail at that format or there's something wrong with your printer's ink feeding or head voltage. It could be a slightly lower ink pressure due to a leaking cart air chamber membrane that isn't reported by the air pressure sensor. Filling the head ink buffers continually but not achieving a real fill where the ink buffer sensors will block the valves again till the buffer(s) get lower again, valves opened and new ink flows to the buffer(s). Ink feeding pressure shouldn't influence the piezo head capacity, it is just used to bring enough ink from the carts to the buffers. Membranes on the ink buffers keep the ink there at atmospheric pressure. But buffers that do not get enough ink will influence the head capacity at some point, bidirectional movement and/or printing speed is the most demanding task. I also wonder whether fine detail shifts the head output to higher frequency + minimum droplet sizes instead of using an average bigger droplet in the other modes. On thin ice here as that is how the 10600 worked.
It could be solved as simple as inserting a new cart. But which one?
met vriendelijke groeten, Ernst Dinkla