First: Obsolete cars I do not consider for balancing, as they are not visible to AIs and new players by default.
This is silly, because if a player starts in 1870 (for instance), many vehicles are not obsolete. If you then play through over a long time frame, you find that your now-"obsolete" convoys are more profitable than potential replacement convoys. In real life, old vehicles keep getting used as long as they're more profitable than alternatives -- VIA Rail uses passenger cars from the 1950s as its main fleet today, and I see rusty old open coal cars rattling by my house regularly. You're supposed to be able to play through from 1890 to 2020, right? New players are going to say ' "wait, where's the car I was just using for that line, it's more profitable than the new one... oh, there is is, "show obsolete" '. That's what I did.
That said, it does sound like you are trying to make obsolete convoys less profitable than new ones, which is the right thing to do. And you are *largely* succeeding.
Regarding coal, all that really needs to be done is increase the running costs of the old_open_car, which is simply outrageously cheap, even in 1875 when it becomes obsolete. Speedbonus is only 2% for most bulk goods, although the food addon has grain (which does benefit from faster travel). Therefore the difference between running costs of .02 per ton-mile and .07778 (for the next available bulk car) is really substantial. You aren't going to make that up on a 2% speed bonus, not for a long time anyway. By 1926 the difference in revenue between running at 45mph and 90mph is only .05 for sand and stone, so the old car is more profitable even before considering engine costs, and even if it runs full both directions! For coal and iron, the difference is .07; that would be an improvement if the cars ran full both ways, but running empty one way the old car is much more profitable. In 1872 when the new coal car is brand new, the situation is only a little better; running empty when returning, the old car is more profitable for everything but grain, though the new one is more profitable if you can run both ways full.
Also the balancing is done for convois that can reach its topspeed. Anything else is completely useless, as comapnies (apart from wild wild west) did never overload their engines or else they would easily stranded somewhere and hinder all traffic.
As an *American*, with the Wild West not far away, I am quite used to companies which did overload their engines. :-) Or more accurately, which used big powerful engines to haul huge trains of slow cargo, rather than to run at top theoretical engine speed.
But for the most part in Simutrans it is only profitable to overload it by a tiny bit -- a train with a top speed of 110 will run more profitably if you overload it so that the top speed is 107. So this is well balanced. Giant bulk coal/sand/stone trains are often more profitable than shorter ones with smaller engines, but this is arguably desirable.
And still by using this scheme, you can earn more money with modern cars, if you can keep them at average 50% load.
Not always. The situation with freight depends very much on which good you're looking at. *Most* of them are pretty good, at least for the rail vehicles, with the exception of the "old_open_car". The more general problem lies in the difference between different modes of transport -- certain goods are far more profitable to run by truck than rail or vice versa, and while that's fine if it's intentional, there doesn't seem to be any logic to it -- at least before 1950, plastics & cement, wood products, and paper should always go by road (which feels especially wrong for wood products, which are often transported by rail or water), and all boxed goods should always go by rail (which also feels wrong). Oil should go by rail (rather than road, fine, but why is rail preferable to a tanker?).
The situation with passenger vehicles is a lot worse, probably because there are more of them, meaning more things to balance. The trams really should be checked against the trains, because giant interurban trams are often more profitable (certainly in every year when a new tram is introduced) than trains along the same route; they can carry a large number of people, the costs per passenger-mile are miniscule, and they make more revenue thanks to the speedbonus structure. (Perhaps this is intended? I know the US actually had a lot of giant interurban trams.) I have to upgrade to trains eventually when the volumes start overwhelming the track, but it causes a sharp drop in profit margins.
This is done by a combination of capacity and speed bonus (if there).
This is indeed how it should be done, but it's not always balanced to actually do this.
There are also some ggods, which are not profitable, but are required to run the whole chain (the AAC example.) This is fully intended.
While there is nothing wrong with this in *theory*, the other elements of the AAC chain are barely profitable too. It's not worth running the chain at *all* unless you need it to make a city grow. If any one step in the chain were fairly profitable it would be worth it.
Although it makes little difference in strategic terms, in terms of "feeling right", I think the *final* element in the chain ought to be profitable, while the earlier ones might not be. In terms of "realistic feel", this is how "industrial development" works: you set up a loss-making line delivering raw materials so you can sell finished goods. You don't set up a profitable raw materials line in order to lose money selling the finished product, which is what's going on with the AAC chain. It would feel a lot more right if AAC was profitable and sand and stone weren't profitable. (Sand and stone are barely profitable anyway.)
Back to the Br39, in 1924 at introduction of the fast passenger cars and the the Br03, you can earn 16ct per large passenger car with Br03 and 7.2 ct with the old cars (at 100km/h). (New cars and Br39 gives less than 11ct and allows only 5 cars, so this is out of question.) The Br39 can draw 6 old cars, but the Br03 can draw 6 new cars.) The maintaince cost of the Br39 is only 18ct lower, which is only a single new car. (Apart from being able to move twice as much people.)
I am going to try to redo your example, but I can't find any "BR03" engine in the game. Perhaps this is the problem? :-) I'll assume BR03 is another name for the BR01...
Example from 1926, since the BR01 isn't introduced until Nov 1925:
(the BR39 seems to have been renamed BR289....)
BR289 + 7 SingCo cars, max speed 97 (same passenger income as 100): running cost 16.08, revenue .19 per passenger * 336 passengers = 63.84 -- net is 47.76 per mile. Or .1421428571
per *passenger mile*. At 50% load, revenue is 31.92, net is 15.84 per mile, or .0942857142
per passenger mile. At 80% load, revenue is 51.072, net is 34.992 per mile, or .1301785714
per passenger mile.
BR01 + 6 HTrans cars, max speed 129 (same passenger income as 130); running cost 78.91, revenue .28 per passenger * 630 passengers = 176.40 -- net is 97.49 per mile. Or .1547460317
per passenger mile. At 50% load, revenue is 88.20, net is 9.29 per mile, .0294920634 per passenger mile. At 80% load, revenue is 141.12, net 62.21 per mile, .1234325396 per passenger mile.
OK, that was surprising to me. Yes, the BR01 is more profitable for passengers -- if you can keep it completely full, it runs in a straight line, and you don't run mail. (Note that this doesn't work with mail -- the volumes will *never* reliably fill up even one large mail car, at least not before the computer slows to a crawl. If you run the mail cars as "economic development loss leaders" on the front of passenger trains, the BR39 train is more profitable.)
So why do I always lose money when I replace 36 BR39 trains on a line with a smaller number of BR01 trains with the large carriages?
I did some more computation and the breakeven point for all-passenger trains, where the BR01 starts being more profitable, is 86% full. It is of course harder to maintain consistently high load factors with fewer larger trains than with more smaller ones. Passengers tend to come slightly irregularly, and passenger loads are usually uneven along the line, and so I can usually achieve an average load of 90% with careful handling, but not 100%.
One extra, and very important, point is that the train route is never as-the-crow-flies, so if you have train-miles slightly longer than earned passenger-miles, you'll find that the breakeven point is at an even higher load. When the train trip is about 16.28% longer than the as-the-crow-flies trip between stations, the BR39 becomes more profitable even for completely full loads. The best you can actually *get* on some routes is 7.96% longer than as-the-crow-flies, because tracks can only go in the eight cardinal directions -- in this situation, the load at which the BR01 is more profitable is 93%. This level of load is actually very tricky to achieve reliably. (If you're playing with pay_for_total_distance=1 or pay_for_total_distance=2 this is even more noticeable.)
So the BR289 is a more reliable profit generator than the BR01, even for passengers.
Which matches my experience. Except for the 221A (which has huge capital costs) the other engines aren't even in contention, until the NS-1000 shows up. Upgrading to the BR01 is usually a serious hit to my profits, and I only do it when the line is simply too crowded to take any more BR39s.
(If I want to overload engines, I would use the Br38, which is even cheaper with only 4.4ct.)
You're right, it's the cheapest in cents per kilowatt-mile. For BR289 this is .0055404958, for BR38 .0050806451, for Garratt .0070344827.
However, unlike the BR38, the BR39/BR289 can "almost" maintain speed with a huge consist. It can stay at speed with 6 of the 90km/h oil cars, and is only down to 72km/h with 13 of them. Running at 72km/h with a BR38 would require two trains of 6 cars each. I ran a lot of scenarios with most of the goods in the game (not mail or trash, and not the ones where road traffic is far more profitable). Prior to 1948 (when the NS1000 shows up) the only times a different engine was more profitable for freights was for Printer's Ink with the 12% speedbonus (which needs to go as fast as possible), and for 45mph bulk trains with the old_open_car, which are cheaper to run with the BRV36.