Electrolysis has nothing to do with thermodynamics
By that, you actually mean everything has to do with thermodynamics. You're adding energy to disassociate a molecule. Thermodynamics dictates that you cannot recover that same amount of energy by letting the constituent elements recombine. Tyr07's uncertain belief coincides with one of the fundamental principles at play in any real world system.
Fuel cell efficiency varies greatly, it goes up to 85% for current marketed high temperature hydrogen fuel cells.
No it doesn't. The only way you could hope to achieve anywhere near that is through some combined cycle process that scavenges waste heat from the fuel cell. You might find some experimental units pushing 70%, but anything commercially available is going to be under 60%.
Total cycle efficiency is going to be under 25%.
That is complete nonsense, you must be bad in math.
Assuming realistic values for electrolysis and fuel cells, you're already well under 40%. Depending on your compression ratio, you're only likely to recover 50-60% of the energy spent compressing the hydrogen for storage, so that's either higher losses, or higher capital costs for storage volume. Tack on a couple percent for leakage, and 25% is very reasonable.