Station Turnaround Time depends on year

[RulerofRails]

I was doing some playing and testing on modern maps. After awhile it dawned on me that station turnaround was significantly quicker than in the earlier time periods. So I ran a test with 8 cars freight in various years with the game set on slow time to look for the trend.

I included all the data I gathered to make it clear that Loading time is very stable, but Unloading time is quite variable, to the point that a small test wont result in highly accurate data. You will notice some years have less data. I increased the number of samples once I realized how variable this value can be.

The trend however is clear: the period 1850 to 2000 sees a steady decrease in time for a station stop.

[Gumboots]

Nice find. Given the lumpy look of the graph, I'd say there's still some noise in the data (which you'd expect). The third data point from the left definitely looks aberrant. The overall trend is unmistakeable though. My bet is the curve that fits it will probably be a basic parabola, like the curve for reliability. So it'd start horizontal at 1830 then curve down to 2000. My guess is it's stable past 2000.

[RulerofRails]

I re-ran the third and fourth data sets (1870 and 1900).

Looks a lot better now. Half the problem was that 1900 was one of my first runs with a smaller sample.

I compared the ratio of Load/Unload time amongst all the data (1850-2000) and took the average (Load is 80% of Unload). This will give a linear approximation. Looks like this. It's a touch low on the approximation side, but this is better than too high for mileage estimates. Also, I tested with 8 cars. I'm almost sure (didn't test for this) that a train will load quicker if you are using a Caboose or Diner on the end of the train.

Load time is stable before 1850 and after 2000. For approximation purposes and because not that much of the game is played in those ranges, I'm going to assume Unload is a constant as well at the same ratio to Load during those periods. Not perfect, but good enough for my purposes.

[Gumboots]

Handy stuff to know. It really would be worth checking if a four car consist loads in half the time of an eight car consist, and what effect this has (if any) on total turnaround time at each station.

[RulerofRails]

I had the same thought, doing a simple check of Load time since it doesn't have the variation and I'm already making the assumption that the ratio between load/unload is approximately the same throughout the game. I also checked did a run with a Caboose, it doesn't affect load time as you would expect.

I gathered data on load time in 1830. Looks like this:

Partial Consist Turnaround.jpg (9.56 KiB) Viewed 4829 times

Bear in mind this is Load Time ONLY to keep the testing as simple as possible. I am assuming that the efficiency ratio of the combined Unload+Load time is the same. I did gather one or two data points that agreed with this assumption.

Pretty clever from the devs. 5 cars and above is within the ballpark, while short consist are noticeably less efficient.

I will point out that this isn't that significant a factor in clogging stations. In sensible play it's not easy to hit the game's limit of how many engines can load/unload from a station at once. Whether Long trains or Short trains can make better use of high density track space is an argument that I don't feel qualified to weigh in on right now.

[Gumboots]

Sure, but this is interesting anyway. If there's a range of high value and low value cargoes at a station you could use this information. As an example, if there is say 4 cars of meat or milk at a good price, and at a fast rot time of course, along with four cars of coal at a low price, it may make sense to deliberately set a train to just grab the four fast-rotting cars and move them out quickly, and leave the coal for a later train. Faster turnaround, along with faster trip time due to higher speed, could make this profitable in some situations.

[undertoad]

I gathered data on load time in 1830. Looks like this:
Partial Consist Turnaround.jpg

Very interesting, thanks for doing this!

Anecdotally, I've noticed the same. I often use "floating" trains without a stable route, which I use to watch out for momentary opportunistic cargos. When one of them is approaching a station, I figure out what can be hauled from there to where for maximum $, change their subsequent routing to go to the right place (don't do this before they start unloading if they have any Express cargo, or a bug will zero your profit on the Express cargoes!) and tell them to load exactly that and how much of it. They load like lightning and get away: especially when loco tech has advanced enough for the consist to be 6,7 or 8 cars.

I'm guessing that your "number of cars" parameter in the experiment is set by setting the train's maximum consist for that station? Rather than the far more difficult (probably impossible) alternative: setting the max consist to 8, but tweaking the cargo available to haul at a profit from the station, so that only 1,2...8 cars actually load?

What I'm wondering is whether the "number of cars" parameter matches what actually gets loaded. Do your fast-loading 8-car-consist train actually load 8 cars, or are they faster to load even if they end up only loading e.g. 4 cars?

It's of interest to me because I hate waiting for trains to load. In a scenario or campaign where there's an offer of "decrease load/unload time in exchange for ?", I always jump at it. I wish there was an upgrade to stations (e.g. an increase to running costs) to allow this. Along with an upgrade to more than 2 tracks, but that's another subject....

[RulerofRails]

What I'm wondering is whether the "number of cars" parameter matches what actually gets loaded. Do your fast-loading 8-car-consist train actually load 8 cars, or are they faster to load even if they end up only loading e.g. 4 cars?

My tests were done with a maximum set using a specific cargo type on Auto Consist. In my test environment there was enough cargo to fill the whole train. Right now I tried a situation where I set an Auto Consist using a specific cargo type with a higher number of maximum cars than were available. I set the maximum as 8, but only 3 were available. Result: took the same amount of time to load as predicted by the little table above: 23 slow-time minutes.

There are special cases for example when a train gets new profitable cargo added to an Auto Consist part-way through a load. I have no idea how that is calculated. I'm a little curious, but I don't see enough benefit to warrant testing time since it's not a predictable situation.

[undertoad]

Right now I tried a situation where I set an Auto Consist using a specific cargo type with a higher number of maximum cars than were available. I set the maximum as 8, but only 3 were available. Result: took the same amount of time to load as predicted by the little table above: 23 slow-time minutes.

Very interesting - thanks for trying this out!

That answers my question. The time taken to load increases with the number of cars - but the efficiency per-car (that you calculated: Time to load / Number of Cars) also increases. And (this was my question), the crucial parameter into the function is not the consist's MaxCars, but how many cars actually load.

So setting higher MaxCars on consists will be of no help in getting trains loaded more efficiently, unless they actually load (MaxCars) cars. The good effect I saw on my "floater" trains must have been because I didn't just give them a big capacity: I also routed them and set custom consists so that they actually did load that number of cars.

Now if we could sort that "train never quite arrives" bug, stations would be far less infuriating! (I mean the poor sod of a train that's crawling along the platform for months at 1mph, trying to come to a stop, and continually being interrupted by other trains swooping in from behind. What's wrong with "first come first served?" Oh for the source code! )