See:
http://forums.auran.com/trainz/show...oute-limits-TS12-and-T2&p=1268142#post1268142
I have been asking this same question to many people here on the Forum.
Sure you can design a 2 baseboard wide route 1000 miles long, but load it up with 8700 railcars, 120 locomotives, 50 genus of twees, 50 genus of schwubbry's and grass's, 500 different textures, 5000 buildings (some designed with high poly sketchup), high poly track, and other splines, 1000's of signals (some of which will drop framerates by 5 FPS when in view) ... And your PC is struggling to decipher, how many railcars and locos are moving, or standing still, and exactly where they all are, even though they are out of the FOV, 50 miles away ... Trainz still feels all these multitude of assets, and is constantly checking each and every one of them every millionth of a second, not to mention calculating all the heights of billions of the 4 corners of each and every 10 m baseboard grid square ... and is choosing what to display just in the Field of View.
On a DEM, I would guess that it is best to keep a route under 100 miles long (is a route is hugely wide, and has 10,000+ baseboards, perhaps it would be better kept down to 50 miles in all directions).
You can do what you want on your own liquid Nitrogen cooled, 4.3GHz, $1000 video card, desktop super computer ... but if a user with a lower end laptop user downloads your route, Trainz may crash to his desktop.
The best checking device is loading your route, or designing your route on a lower end laptop ... it it runs good on an old beater, clunker, laptop ... it will run even better on a high spec PC.
If I were doing a series of routes I would break them down into many, many iPortal, or Portal segmernts.
ie:
NY to Phila
Phila to DC
Phila to Reading
Phila to Harrisburg
Harrisburg to Williamsport
Williamsport to Tyrone
Harrisburg / Enola to Mount Union
Enola to Harve DeGrace
Mount Union to Johnstown
Mount Union via the EBT RR and Robertsdale
Johnstown to Pitrcairn
Pitcairn to Conway