First thing I discovered is that unlike fuel consumption it's not initially dependent on weight. All brand new locos with the same reliability rating will have the same breakdown chance.
Next thing I figured out is that the breakdown chance is very badly worded. I'd always just assumed it was what it said: a percentage. In other words, 15% is 15 out of 100. I just noticed
that the text that shows when you hover your cursor over the breakdown bar says it is a chance "out of one thousand miles". This is a new and exciting definition of "percentage". Percentage has apparently been redefined so that 15% means 15 out of 1,000.This, although colossally idiotic, would explain why trains don't break down as often as their breakdown bar suggests they would. If the breakdown bar says 25%, what the stupid thing actually means is 2.5%. Things like this make me contemplate the usage of my cornucopia of naughty words. Anyway, just take the "percentage" and divide by ten if you want an idea of how often your train is likely to break down.
Moving right along, empty brand new locos sitting at a station are all very well, but you need to load them up and use them. So I did. A brand new unloaded J class has a breakdown chance of 1.3 not-percent. Load up the J class and it jumps to 3.9 not-percent, or three times the unloaded value. For comparison, a new unloaded Kriegslok rated Above Average for reliability has a breakdown chance of 3.0 not-percent, which jumps to 9.7% not-percent with 8 freight cars, or a bit over three times the initial value.
Ok, so this is the for new locos. Next thing is to find out what happens with age. For this I just left locos sitting empty at the stations, and kept track of how their breakdown chance changed. It doesn't change during each year. It only changes at the start of every new year, at the same time as the displayed age of the loco changes. What's more, it doesn't always change at a constant rate. It rapidly gets worse in years 0 to 2, increasing by a factor of around 3.5 depending on reliability rating. After that it slows right down and changes to a constant rate. This means your locomotives are almost as unreliable after 3 years as they are after 8 years.
N&W J class goes 1.3 > 2.6 > 4.5 > 4.6 > 4.7 > 4.8 etc, at 0.1 intervals after the first few years.
The Krieglsok goes 3.0 > 6.2 > 10.8 > 11.1 > 11.4 etc, at 0.3 intervals after the first few years.
The old Firefly goes 8.6 > 17.9 > 30.7 > 31.7 etc, at 1.0 intervals after the first few years.
The change in breakdown chance when loaded tracks the increased unloaded chance pretty well.
J class at 9 years old has a breakdown chance of 5.5 unloaded, and 17.2 loaded. This is still close to three times, just like it was when new. A 9 year old Kriegslok is 13.1 unloaded, and 42.0 loaded. That's a bit over three times, same as when new. The 9 year old Firefly is much the same. It's 37.4 unloaded and a massive 125.8 loaded, which is a change of 3.36 to 1. Its ratio is about 3.57 to 1 when new, so pretty much in line with the J class and Kriegslok at the same age.
Now I just happened to have looked at a saved game of mine that was running a pile of Kriegsloks on the Italy map. A fifteen year old Kriegslok that had just been sitting at the station would have a breakdown chance of about 15 unloaded. Loaded up it would be about 48. Funnily enough, a fifteen year old Kriegslok that had been hauling freight up and down the coast from Rome to Salerno and back had a breakdown chance (when full of oil) of about 24. Since this train was running 7 D era freight cars plus a caboose its breakdown chance was halved, so that's in line with the 48 you'd expect without the caboose.
"Full of oil" is the next part. By tracking that train and several others over the course of their runs, I figured out that breakdown chance increases by 1 not-percent for every 5% decrease in oil level. Again, this was with a caboose, so without a caboose it'd be increasing by 2 instead of 1.
IOW, if you aren't running a caboose, and if you are relying on inline sheds that won't even look at your locos until their oil is less than 50%, and assuming your loco is rated Above Average reliability, then by the time your loco stops for oil its breakdown chance will be at least 20 points higher than just after it has filled up with oil.
I was servicing mine on spurs, so their oil never dropped below 70%, and of course they had the caboose as well. That means their reliability never dropped more than 3 points due to oil over the course of their service lives.
So that's the basics for reliability changes with age, oil level, and load, on terrain that is close to flat. I haven't had time to do an in-depth study to see if grades make any difference. I'm starting to suspect they won't. I am getting the impression it's only dependent on age and load (yes RoR, you did suggest this
).After all that, what does it mean for our cunning plan to increase freight weights? My 2c is that I know Kriegsloks with 7 freight cars and a caboose will give decent post-1950 service for up to about 15 years on fairly flat terrain. After that they start getting likely to crash and burn, and their maintenance costs are getting high too. On moderate grades I usually load them with 6 cars and a caboose.
I can use that as a guideline for changes in freight weights. It comes down to deciding how long we want locos to be reliable enough, and economical enough to run. If we want a lifespan up to about 15 years, a Kriegslok will do that hauling 840 tons of freight and a 53 ton caboose. If we want to increase the consist weight by 50%, then the loco will need a reliability rating that will handle that load with about the same breakdown chance as a Kreigslok with an 893 ton load. If the loco hauling the heavier load doesn't have a caboose, its loaded breakdown chance will still have to match the Kreigslok, with caboose, to give the same useful service.
In rough terms, a J class hauling 8 freight with no caboose should have about the same in-service reliability as a Kriegslok hauling 7 freight and a caboose. To put it another way, in terms of breakdowns a caboose is worth roughly two levels on the reliability scale.
Once we get a few more figures to narrow down exactly how consist weight affects breakdown chance for varying loads, this should all sort itself out fairly quickly. It's going to be a fairly tightly constrained solution. We'll know for sure how high the maximum freight weights can be, and then just work backwards for mixed traffic and express locos.
So, this info seems useless. On the bright side, at least it's not another place where the game is broken. An oddity.
.