Something always seemed off when applying the concepts in this thread to my commercial designs. Smaller propulsion plants consistently looked better than the supposed optimum. Apparently I'm an idiot, the problem was the underlying assumption that speed is relevant in itself.
That assumption is fine for how I design warships. Too slow and we can't control an engagement, doubling our firepower may not be enough to compensate. Too weak and our speed is only useful for running away, leaving behind shattered fleets and glassed worlds.
However, When the expected service is hauling goods or colonists year after year, speed and capacity are interchangeable.
Assume we have a given tonnage available for engines or cargo capacity. Defining x as proportion of engine tonnage, our throughput is proportional to x(1-x); 4x(1-x) if standardising for the maximum, achieved at x=0.5. Using a given type of engine, our relative fuel efficiency is (1-x): A colony ship with 3 size-50 engines and 1 cryogenic storage has the same throughput as the reverse, but needs to make 3 trips instead of one, consuming 3 times as much fuel.
If we can freely adjust our power multiplier, throughput scales linearly with it while fuel use scales with a power of 2.5. Therefore, we can achieve the best fuel efficiency by maximising (x(1-x))^2.5*(1-x), achieved at x=5/12 or 41.67%; deviating from this is less fuel-efficient than adjusting power multiplier. Standardised for 1 as the optimum: (x*(1-x))^2.5*(1-x)/0.01699
Of course, we don't have to stick to this slavishly. I may choose less tonnage in engines if I don't have ultra-low-power engines researched yet but still want low fuel use, or when even our lowest-power engines are expensive compared to the payload - quite likely for freighters. If we can't adjust power multiplier downwards, we may simply look at throughput*fuel-efficiency: (x(1-x)*(1-x)/0.148148148 .
Similarly, I may choose more tonnage in engines if I can spare a little fuel and am stuck at the 0.5 limit for commercial engines. Or when the ship is held in reserve for sudden priority use, rather than continuous service. Or when the payload is much more expensive per ton than engines of the optimal fuel efficiency (likely for colony ships).
All graphs in one: 1) throughput*fuel efficiency with fixed-power engines, 2) unadjusted throughput, 3) throughput at constant fuel efficiency with freely scaleable engines:
x(1-x)*(1-x)/0.148148148, 4(x)(1-x), ((x)(1-x))^(5/2)*(1-x)/0.01699