I would like to inject the idea that MODELING freight train consists on layouts
is not precisely the same thing as understanding what a particular prototype
may have done at a particular point in time...
In my opinion, if your layout represents a mainline railroad, you are going to
have a particular mix of model cars on your layout and you will need staging
and you will need to have train consists that
(1) represent 'possible' or reasonably realistic consists given your location and
time period and scenarios
(2) are variable, so that trains do not all look exactly alike and may even cause
an occasional incredulous 'Wow, I didn't expect that' but on reflection you
may agree surprises are part of the real world experience of train watching
Ten years ago or so I fooled around with software that combined an amount
of randomness coupled with a number of 'rules' and I ran thousands of train
consists simulating dozens of years of operations at the North Shore RR club.
Every now and then I would get really interesting results, but generally train
consists were highly consistent with their purpose and types of cars and the
road names and destinations of cars.
Unfortunately, layouts that use car cards are not amenable to this approach,
unless you want to deal with printing new waybills all the time. Car cards are
very labor intensive if you want to operate realistic and variable consists for
hundreds of cars. Also, variability means you have to have extra storage or
staging to handle all those 'occasional' cars.
On a private layout, you can control the exact mix of equipment according to
the fleet statistics carefully compiled by Tim Gilbert and other folks that have
made studies of conductors' books and interchange records. But personally I
think a layout that completely lacks the occasional oddball car is diminished if
you operate it a lot, because it starts to seem very predictable and routine.
On the other hand, rules for oddball cars should ensure that they do not show
up all the time. Some may only be seen once a year.