I have fun making these schedules and I think I get pretty close to what can reasonably accomplished, but I’m also well aware it’s an approximation.  There is more involved in crafting a schedule than figuring out how long it takes to get somewhere.

Of course trip time is the starting point.  From observing other schedules I’ve worked out some rules of thumb regarding timing:

59 mph track averages 1.2 minutes a mile speed.
79 mph track averages a mile a minute speed.

And so on, figuring between achievable speeds of between 80-82.5% of theoretical speeds.

These speeds that I use are slower than theoretical speeds because a margin needs to be considered to allow for slow curves and bridges, etc.  In many cases, achievable speeds are higher than these rules of thumb (which I know because timetabled speeds are or once were higher).  In other cases they could be lower, but this just points out speed restrictions that should be lifted if the funds are available.  I have available to me timetables for all rail lines in the US, Canada and Mexico and use them (I have official guides from 1920, 1953, and 1967 and Amtrak timetables from the seventies, eighties, nineties and since then).

So when I make my schedules I’m following a scenario with hypothetical constraints, something like “what would service be like if track was improved to 59mph standards,” and so on.  I always know the speed limit of the line and make the timetable accordingly.

What I don’t know are all the details.  There are so lines I know well enough to know all the speed restrictions, but mostly I just go with my rules of thumb, which is necessarily imprecise.  Of course the big ones I do know, even in places far away, and I do account for those, putting in extra minutes for known delays.

And then there is acceleration and deceleration.  For an average passenger train I’ve worked out these rules of thumb for minutes to add for each station or slow order: 1.5 minutes at 59mph, 2 minutes at 79 mph.  4 minutes at 125 mph (based on metroliner and Acela timings), 6 minutes at 186 mph (based on TGV timings).  Dmu’s have better acceleration and passenger trains with heavy mail and express loads have worse, but I don’t really know how much so.   Dwell time is a function of how significant the stop is.  Small stations have 30 seconds or less.  The TGV dwells 2 minutes at Macon, a town of similar size to Burlington VT.  My opinion is that Amtrak dwell times are far too lengthy and could be tightened considerably if passengers were cued to be at specific boarding locations ahead of time and if double and triple boarding was eliminated by extending platforms and if passengers were on the platform ready to board BEFORE the train arrives, rather than after alighting passengers have left, as is the practice at some big terminals.

I like to add stations where I think Amtrak is skipping a potential market, usually suburban locations on interstates or with high populations.  My rule of thumb is to stop where there is a population of 50,000 or more to be served – 100,000 or more if it’s an express train.  Of course the lack of competitive alternatives makes a huge difference – if there is no nearby interstate and no nearby airport, the train will do better there.  I think Amtrak underestimates how much boost to ridership it could get from convenient stations, even in regional markets already served by a central station.  Sometimes I’ll also add a station where I know the train already has a slow order (meaning that stopping adds only seconds to the schedule).

Another factor, very important, is congestion.  Mostly I duck the issue and make assumptions that my timetables are for service where money has been spent to take car of these issues.  On single track lines schedules have to be timed for meets at sidings.  Mostly I don't know the locations of sidings, so I duck this issue too.  Closer to home, I'm able to take this into account.