I've been working on quite a few prototype-based custom track plans lately. I enjoy these projects, but it's always a matter of making trade-offs between what was actually there on the real-life railroad and what we have room for on the model. There are sort of three ways to approach the challenge.
The first way was used a lot in the 1950s and 1960s and can be seen in older plans such as those in Kalmbach's 101 Track Plans. Want Newark, New Jersey on your HO layout in ten linear feet? No problem: just put in a siding, two stub-end yard tracks, a couple of industry spurs, and voila!, it's Newark!
Of course, this is barely adequate for Newark, TX (on the Rock Island), let alone Newark, NJ. Yet this caricature style of "prototype" track planning persisted through even some well-regarded published plans -- and widely in general use. These highly abridged scenes didn't look very much nor work very much like the real thing, but it did allow the designer to claim 300 miles of the prototype in a spare bedroom.
The other extreme has come into vogue recently with the wider availability of prototype information. That's to simply scale down the prototype element by whatever ratio necessary to fit a copy into the allotted space. The real yard is two miles long and you have 15 feet in HO? No problem: just shrink the real thing by a factor of eight. So what if it leaves you with body tracks three inches long? It's an LDE, don't cha know. What could be more accurate than that?
Of course, the best approach requires a lot more thought, and that's why it sadly eludes so many designers. We have to decide what balance of "looks like" and "works like" best fits the available space and layout concept. And then work toward capturing signature elements that suggest the real thing.
On a project I am completing now, one of the challenges is a scene that was about half of a mile long in real life. It needs to fit in about 600 scale feet of benchwork between two curves. While a simple 4-to-1 compression would theoretically fit, it wouldn’t capture the personality of the signature elements, which include a truly massive industry, a very modelgenic station, and a couple of smaller typical Midwestern rail-served businesses.
Fitting it all in while including a bit of the street grid that helped define the real scene required flipping one spur to point west instead of east and placing a station on a curve. This allowed the track configurations around the large plant to more strongly resemble their real life counterparts.
One of the keys is modulating the degree of prototype fidelity around the typical layout. Signature scenes in key locales get more focus (i.e., space), while others are more heavily modified. This process requires a firm grasp of the overall concept and vision, which then serves as a reference against which to judge the many trade-offs required.
Borrowing from W. Allen McClelland's "good enough" modeling motto, this "close enough" approach allows us to capture the most appealing and engaging scenes and elements in a way that communicates the atmosphere of the real thing to viewers and operators. Modulating fidelity and referring back often to the guiding layout vision helps create realistic scenes in reasonable spaces.