General management strategies

[Gumboots]

My standard consist for just about any scenario or period, for locos which have moderate reliability and on good terrain, is 6 cars + caboose. 5 cars for the very early locos, or on worse terrain. I prefer to keep the number of trains lower and get good haulage out of them.

[RulerofRails]

Mostly I haul 8 cars or 7 with a caboose when reliability is Average or less. Adverse terrain is the most common reason for decreasing this. The arrival of the 2-8-0 Consolidation (2-6-0 Mogul in 1.06) makes full trains my default preference. Before this I will go for 6 cars as the default. Of course there are special circumstances such as speed goals and I tend to choose the engines with the best pulling power while weighing running costs, reliability, and engine cost. Acceleration is in my view the achilles' heel of engines, I would rather have a longer, slower train that doesn't stop rather than fast ones that stop and start a lot. Double track can alleviate this, but this is an extra cost to add to that of the extra engines that will give you a poorer return on investment for rails, which is in most cases poorer than industry already.

More Subtle Nuances:
Price differentials between stations on the demand map take awhile to build up as well. Sometimes less frequent service will allow a bigger price difference to build up, essentially the cargo has to sit at the station for a reasonable demand to build up. Suppose a load of Produce arrives in a station. It can be hauled away immediately at 2k profit, versus waiting a month or two to get a 10k difference (profit) for that same run. This is why optimizing rail revenues is all about timing (as well as routing). Point here is that frequent service isn't always the best thing.

Also, having a longer train may sometimes mean that all of a cargo is "cleared" from a station. Cleared: all cargo of a particular type is removed from a station. This will allow demand to build up quickly in that station again. (Of course part loads will still linger, but they will likely be consumed and this process can still start.) Suppose (this example assumes you have no new supply, either a producing industry or incoming train in this timeframe) you have 6 cars of Meat in a station, a train with 4 cars will leave 2 behind. This may be picked up by a second train pretty quickly (service frequency is higher), but the profit for those last two cars will start decreasing as soon as there are no longer the full 6 giving an "over-supply" situation. In addition there is a delay before the station is "cleared" causing a delay in possible demand recovery. Short of this one: Most profit (use of demand gradient differential) is to be had from hauling all of a particular type of cargo in bulk.

I am not pedantic about this stuff and don't try to use it for micromanaging even though that is very profitable, but oh so tedious! However, I try to work these principles into my train routing in some way. Especially: when I expand and how I set up new routes in my rail network. When a map has developed resources it is fairly easy to match industry ROI over the short term, here I am trying to continue that difference.

[Gumboots]

I did do quite a bit of testing with 7 cars and caboose, but found it had a noticeable and adverse effect on breakdowns and crashes compared to 6 cars and caboose. Moving back down to 5 and caboose didn't seem to be more reliable than 6. That's why I settled on 6 as my default. It gives good haulage returns without ramping up breakdowns.