Size of Drivers on a steam locomotive. - Trains Magazine

Wide variety of issues involved with driver size. Some more complexity is introduced because the mechanical stroke (of the piston) determines the crank circle on a given locomotive.

Remember that (in addition to the valid points made by beaulieu et al.) you have balance issues with rodwork, wear issues with driver tires, suspension, equalization, and balance considerations (including rigid wheelbase), brake-rigging issues, effective frame height and boiler packaging within a given loading gauge ... just to name a few off the top of my head. In a two-cylinder double-acting locomotive, there are relatively 'hard' restrictions on the power available at or close to starting that are related to driver diameter, and the smoothness with which that power can be applied at the railhead and to the trailing load.

It's easy for 'armchair' locomotive designers to start talking about running small-drivered locomotives at higher rps because modern materials and balancing technique (and advanced valve gear) would permit it, but ignore the relative effect of, say, induced pin failure at a given track speed, or take the trouble to calculate whether rotational inertia (and various augment forces) increases or decreases vs. a larger-diameter alternative for a given range of speed. On the other hand, it's easy to misunderstand whether 'high speed' is even an effectively important design criterion for a particular application...

Of course, what you see 'historically' is very often justified for reasons different from straight theoretically-based engineering. Sometimes this is good, sometimes bad. One point made last year in the Trains article is that very often steam locomotives spend much of their working life running at much slower speeds than 'design' would otherwise warrant -- it may therefore be valuable to design them so that higher speed is possible (but not optimal) but the overall 'cost of ownership' is minimized in reality -- which imho is a good and reasonable criterion for locomotive design. If you intend the thing to run at high speed for much of its life -- cf. the Milwaukee class A's, which weren't really suited for much other than that, or the ATSF 4-8-4s -- then you are justified in using larger drivers and tinkering with stuff like multispeed boosters to optimize the main locomotive to run most efficiently. But beware the wrong assumptions -- the Alleghenies and the PRR divided-drive freight power being two important miscalls (the former because the only service that could cost-effectively use their single-unit power ran at speeds well below their best developed power, and the latter because PRR freight speed limits forced the same condition!)

As an interesting exercise, you might consider whether an express engine on most American passenger service (in the days of steam) *required* drivers higher than the 70" non-disc variety found on the N&W J class. My own opinion (not worth much, necessarily) is that the 'sweet spot' for normal fast-service power is somewhere between 72 and 77 inches (this won't change if you use precise valve gear like Franklin type B, but more because of suspension and peripheral considerations than rotational speed or steam consumption) and there's little point in going smaller than the high 50s for anything intended to operate on a normal-speed mainline.