Service Facility placement

[RulerofRails]

Following the discussion over here that started off when I forgot about the 1mph bug , I did some tests on service facility placement deciding to start a new thread to prevent off topic there. The majority of my tests revolved around 8-car Iron trains hauled by the American 4-4-0 and 242 A1 as they are in 1.06.

I tested on slow time with flat, perfectly straight track using two setups: Long and Short. Long being as close as possible to the distance that a train can go before running out of water. Short is half of that enabling the train to make a return trip without a water refill.

All "spacing" measurements count the number of clear cells between two facilities counted as the green numbers on the track grade view. Hopefully that's not confusing. For reference, a maintenance shed as close as possible to a Large Station leaves 4 clear track cells. Engine length may play a part in the 1mph bug, so to be really safe I would recommend 6 clear track cells of separation.

With "Service Near Stations", I always had the maintenance shed closer to the station (with identical distance between the station and shed as between the shed and tower). Because the maintenance shed doesn't get used as often, optimal placement puts the water tower closer. However, the way I do is a practical insurance against a rouge long-distance train accelerating with no water.


This shows the how much time is saved by service is placed near the station as a percentage of the mid-route total time.

Mid vs Near Station service - Revised.jpg (11.65 KiB) Viewed 5423 times

There is an issue with my test: the Mid-placed service doesn't have the same service pattern on the short route. Stopping for water twice as often means that it takes on only 51% each water stop. I have revised to correct this. I took the quick route by discounting any time spent filling with water. On the sheet below which remains raw, the Near Station times dropped by 16 and the Mid-route times by 8. Then I calculated new percentages.

It's not a big saving. Hard to justify near the station service in real-world conditions with only this.
ETA: if I had gotten it right the first time I wouldn't have written that.

The differences in time between the 4, 6, or 8 empty track cell counts are so small that I didn't bother to calculate them exactly. Here's my summary sheet for detail-orientated folks. The figures are minutes of slow time. Remember that 60 minutes = 1 normal month.

The 1mph engine bug is only very slightly present at 4 empty track cells. Possibly, engine length contributes but that's unconfirmed for now. As long as the 1mph bug is avoided, closer does seem very slightly better. Traffic constraints should take precedence over any potential gain here for sure. I definitely over-estimated this so will tweak my strategies accordingly.

---

Also did a quick test for the junction bug:
Trains will stop from either direction at service facilities that are 6 empty track cells from a junction.

---

Side-note:
I have enough evidence to be confident in a quirk I have known about for some time. On runs of the specific length between 40%-50% of the water consumption required, "Near Station Service" requires half the number of water stops that service placed right in the middle service does without running out of water. Reason being that more that the whole tank of water is being utilized instead of just 50%. The same effect can be had with service in the countryside if it is offset to be within 20% of the water consumption distance from one of the stations. That way the train will pass by every second time still above the 60% refill level.

A water stop takes a long time, about half a month stopped, then further time when accelerating again. The American was losing a little over 1 month (69 mins or so) per trip that needed an extra water stop. For reference a non-stop trip took just over two months (130 minutes). That stop increases trip time by 52%.

Better acceleration and higher speed for the 242 A1 doesn't help. The extra water stop only takes 3/4 of a month (about 45 mins). But, non-stop time is 66 minutes. Giving a higher ratio of increase at 68%.

This is why I have dabbled with one stop round-trips on routes that are slightly longer than the water tank allows. But, only where traffic is low and the portion to be traveled without water is not uphill.

Edited: because I goofed badly on controlling a variable in the test.

[Gumboots]

It's not a big saving. Hard to justify near the station service in real-world conditions with only this.

Not a big saving on short runs, presumably because the acceleration from the first station is making for a slower average speed anyway (being a greater proportion of the trip on short trips). OTOH the time savings on a long trip are enough to be useful IMO. 6% up to 10% is getting to be substantial, and could add up to a fair amount of extra revenue over time.

The next thing to test would be servicing on spurs, to see if it actually benefits total turnaround time compared to inline servicing. I know it benefits "average speed" but that is not the same as "average number of trips it can fit in a given time". The latter is more important most of the time. The former is only relevant when game goals specifically call for a certain "average speed".

Also did a quick test for the junction bug:
Trains will stop from either direction at service facilities that are 6 empty track cells from a junction.

How about service facilities near to waypoints? Are they affected by the same 6 cell rule?

[RulerofRails]

I edited the OP, because I realized I didn't control the water fill time across the test. It makes more sense now IMO.

Tried a quick test for a service spur on the short route with the American 4-4-0. I stated above that stopping for water-only took 69 mins longer than a no-stop run. This is with the mid-route setup which only contains 8 mins for filling. Whereas, thanks to the correction we know that the Near Station time of 62 mins includes 16 mins for filling. So we use 62 mins for water-only and 117 mins for full service.

I built the most efficient spur by building the facilities on the straight track that exited the station in the other direction. I tried 3 placings and tried Shed closer then Tower closer. Best I could manage was a water-only time of 86 mins and in a different iteration a full service time of 144 mins. Definitely not faster.

Overall, I think this has sunk the ship on spurs for me. I will still use them to trick the average speed goals. BTW, I tried to use average speed for this test as well, but frankly the readings aren't accurate in a repeatable way.

I couldn't get a train to skip service because of a way-point. A Water Tower on what I believe was the neighboring cell worked fine. Do you have an example saved?

[Gumboots]

With the spur test, were you accounting for loading and unloading times? If you were timing station to station with inline facilities then that wouldn't include any loading or unloading time, whereas with spurs you have to arrive at the station and unload before going to the spur.

So the way to do the comparison would be from departure at one station to commencement of loading at the second station. That will determine the actual turnaround times for systems with inline facilities, and systems with facilities on spurs.

I think I have an example of the waypoint problem saved. I'll have a look and see if I can find it.

[RulerofRails]

Didn't need to. None of the figures include station loading or unloading times. I measured only the time between stations.

Stations stops seem to consist of only two parts: Loading and Unloading. Trains without a consist to load will depart as soon as Unloading is complete, and in-bound empty trains will begin to load as soon as they stop. Easy to see the second thing during normal gameplay.

I have news that Unloading time is more random than I thought. Saw 52-125 minutes during testing. Eventually started using -10,000% loading times, which gives instant station stops so I didn't need to make heaps of subtractions. Just to be double-sure, I used the default when testing the spurs.

[Gumboots]

Ok, you've lost me. I looked at the post you linked to and this doesn't make sense:

I took a little time to do some tests on station and service stop times. I put the game in slow mode and then double checked that 1 hour of slow time = 1 month regular time.

The basics are that to load and unload an 8-car train takes about 1 month. Each service stop takes about 1/2 a month. This is the measurable part that should be consistent across all locos. I discounted the time it takes the train to accelerate back to full speed.

Why would you do that if you're trying to test the effect on trip times? I would have thought that was a highly important factor to include without discounting.

Anyway, I know about the loading/unloading pattern at stations. AFAICT the time it takes is dependent on how much cargo is available. If there are only a couple of cars that can be filled, it'll load much faster than if there are a full 8 cars of cargo available. Unloading time seems to be much the same regardless of the number of cars. I haven't tested all of this by timing things. It's just the impression I got by watching what happens during play.

I'm not sure what this means either:

Eventually started using -10,000% loading times, which gives instant station stops so I didn't need to make heaps of subtractions. Just to be double-sure, I used the default when testing the spurs.

Do you mean you tested spurs with instant station stops and with normal ones?

Just to be clear, IMO the important cycle is this:

1/ Train leaves first station with full load. This is t=0, and is when the "Loading" text vanishes from above the train in the train listing.
2/ Train trundles down track, stopping at line facilities if they are there, or otherwise just going straight to the second station.
3/ Train unloads at second station.
4/ If using spurs, train then proceeds through service cycle on spur, otherwise go straight to...
5/ Train loads at second station.
6/ Train leaves second station with full load at t=whatever.

t=whatever is the only number that's relevant, IMO.

[RulerofRails]

Why would you do that if you're trying to test the effect on trip times? I would have thought that was a highly important factor to include without discounting.

The link to the other thread is from some time ago. I only included it for the graph. Over there I was only measuring the amount of time the loco wasn't moving (except for the "Stopping" figure that I shouldn't have included). For the current test it's 100% important, and the reason why exact figures can only be loco-specific.

For the spur test:
Time measurement started the moment that the station stop (which is just an unload) was complete. In my definition that's the moment the train destination displays as the service facility.

The train was then allowed to proceed normally through the service facilities. The clock was stopped the moment that the engine came to rest at the station. Cars are visibly loaded onto the train at precisely the same moment.

All the other measurements were made:
For an 8-car train that has just loaded at Station #1, the clock starts the moment that it displays Station #2 as the destination. The clock stops as soon as the train comes to rest in Station #2 at which time it will display the "(Unloading)" notice.

That's the basis of the Non-stop, Water Only, or Water & Oil figures depending on if the train stops for in-line service. To get the comparison between in-line and spurs, I subtracted the non-stop time from the Water Only or Water & Oil times.

As to your points, #3 (time the "Unloading" message is displayed) isn't a constant according to my measurements across many test with identical trains. I have no idea why. If you want, test if this number is a constant for you.

Do you mean you tested spurs with instant station stops and with normal ones?

Spur test was done only with normal ones.

[Gumboots]

I found that game save I was talking about earlier, but I had remembered it incorrectly. It wasn't bugging because of a waypoint near service facilities. It was a different bug.

What was happening was that I'd built inline facilities on a short side track about halfway between Fort Worth and Memphis on the Blue Streak map. This was for servicing long distance ore trains that were running straight to Memphis. I'd set the service tower on the side track as a destination, but there were other service towers earlier and later on the same general route. These weren't set as destinations because a/ I was lazy and b/ they were on the main route so c/ the train should have used them anyway when it was low on whatever.

The problem is that it didn't use them. It would ignore all earlier facilities and go straight to the one that was set as a destination. This invariably meant it spent quite a lot of time running with no water. It seems that setting a service facility (that is between stations) as a destination borks the train's ability to recognise when it should be stopping for a top up.

The probable way around this, which I'll test tonight, is to change the routing so that the facilities are not set as destinations, and instead use a waypoint on the side track to make sure the train passes all facilities. Based on what you found with your testing, this should work.

[Wolverine@MSU]

That's the bug I have seen too. I'll be interested to learn what you find out about setting a waypoint along the spur track as opposed to setting the service tower itself. I suspect that this will solve the problem. Thanks for checking it out.

[Gumboots]

Yep. it works. Waypoints work perfectly. Train will stop anytime it needs anything.