I had the derailment problem on my Sunset J1a (still one of my favorite engines) when the lead driver's springs would drop the height of the engine so much that on O-88 curves the blind middle driver would fall below the rail head, when it hit the straight, the middle driver, now off the rail, couldn't get back on and would derail. I used the plastic plugs only on the forward axle to keep the engine level even on curves and the problem was solved. Problem is the engine is heavier than the spring rates selected. When all five drivers are supporting the engine, no problemo. But on a curve when the middle driver actually floats out over empty space, AND no longer supporting its share of the engine's weight, the engine would droop. A ten-coupled engine with 63" drivers requires a huge turning radius.
The real engine was made with blind center drivers when first built (like the model) and had lateral motion devices to permit the middle axles to slide sideways to adjust for tighter turns. When they refit tires on the engines they went back to a flanged drivers. I guess they figured that the Pennsy had sufficiently wide curves to handle the engine. Of course, this engine and other long wheel-based beasts like the T-1s, S-1, and Q's, all ran on the Western end. People in Philly never got a chance to see these engines in action.
A Baldwin Centipede actually got stuck on the Zoo curve in West Philly. It was a mess and tied up the main line. Baldwin, unfortunately, made diesel engines like their steam engines with big, unbending, monster castings on their under frames and GM didn't. GM won!