Train yard

[Martijn Meerts]

I've been looking a bit at building a train storage yard recently. Obviously, considering I'm still planning on building a full length Tokyo Station, I'll need a yard that can hold 16-car trains =)

My modules will be 70cm in depth, which means a maximum of 28 tracks. An initial track plan (large image ;)): http://www.jr-chiisai.net/_div/trains_temp/yard_v1.jpg

The yard basically has 2 "track harps" as they call them in German (Gleissharfe or something along those lines, guess it's because the turnouts and track make the thing look somewhat like a harp ;)) The 2 harps are connected with the loop on the right side. The reason the loop is so big, is that the loop in itself needs to be at least as long as the longest train, considering it's actually a return loop.

Each of the harps keeps 1 track open at all times, allowing trains to pass (track 14 of the upper harp, track 28 of the lower harp). Several tracks will be cut up into multiple sections, so that in case of short trains, more than 1 can be stored on a single track. Some tracks will be divided into 2 equal length parts for 8-car trains, some in 2 non-equal length parts for say 10-11 car trains and 5-6 car trains. Others can be divided into 3, 4 or even more parts, depending on the trains that need to be stored.

My problem though, is that the entire construction is just over 5 meters long, which means I'll need to install some new stairs up to the attic before I can place it anywhere =)

Another problem is obviously the amount of turnouts (52 of them if you don't feel like counting). As these turnouts need to be dependable, they'll be driven using servo's. Adding DCC controlled servo's to 52 turnouts will be quite interesting. I can cut it down to 26 turnouts if I decided trains should always run in the same direction, which is likely what I'll end up doing. That way I only need to motorize half of the turnouts. But that still means I need 26 servo's, as well as 7 or 8 servo decoders (1 can control 4 servo's). Turnouts will be Peco (electrofrog), so I might need a way to polarize the frog. I know the ESU servo decoders have an extension module that does exactly that, but it would cost 50 euro per 4 turnouts, not counting the servo's themselves =)

I've considered using stub tracks since a lot of trains can run both ways without problem. It would mean I'd need less space, but stub tracks have their own issues when automating the entire thing. One obvious issue with stub tracks is that it's much more difficult to divide tracks into shorter sections.

One advantage of the design as it is now, is that it works nice for exhibitions. People can see a lot of different trains (if they stay and watch a while, the Peco track is so close to each other, that the only thing you'll really see are the roofs ;)) and the blank space area in between the loops could have some sort of interesting display things.


The worst part though, it that I probably have more trains than would fit on the yard =)

[CaptOblivious]

I have heard that one way to save linear space in a yard, with the same number of turnouts, is to use two layers of turnouts as such: you have one set of turnouts, one fork of which leads to a yard track, and one fork of which leads to another turnout, which then leads to the yard on both forks.

[Martijn Meerts]

I've already done that, it's just not all that obvious because I'm using the medium Peco turnouts. I can use small turnouts and save some space, but the radius on them is smaller than the minimum recommended for shinkansen, and I'm not sure I want to take that risk. (and I don't think it save more space than maybe 10-20 cm)


Edit: I did a quick test. Using Peco's small turnouts (305mm radius, might not work well considering you get the whole s-curve thing when doing a yard like this) I'd save about 30cm on a total length of 510cm. Tracks would also be even close together then with the medium turnouts =)

[railzilla]

Instead of a harp you could use a tree topology. After every turnout you place another one means after 4 turnouts you have 16 tracks. Advantage is that the number of turnouts a train passes is always the same regardless on which track the train is going to. Also the length of the tracks is more uniform.

Why you insist of servos? Using tortoise is as reliable as using servos. And there are DCC decoders which can operate 8 tortoises. This would cut down your cost at the expense that the tortoises are quite bulky.

[Martijn Meerts]

Do you have a sketch or picture of a tree topology? If you mean what I think you mean, there are a couple of problems. It would require 3-way turnouts, which Peco doesn't have in code 55. It'd also be more difficult to get the return loop. But, as I'm not 100% sure what you mean, I might be completely off =)

The problem with the tortoise is the fact that they are so big. I have one that I ordered a long time ago to play around with, but I can fit about 10 servo's into 1 tortoise =) The less space used by servo's might be offset by the additional decoders I'd need, but the Tortoise also requires a rather high clearance.

[omar]

while not peco, have you considered Kato unitrack? you can get power routing and selfcontained electric switching. just a thought.

[Martijn Meerts]

while not peco, have you considered Kato unitrack? you can get power routing and selfcontained electric switching. just a thought.

Unitrack is more expensive, less flexible and not code 55 :)

Space between the tracks would also be bigger, so I would have to lower the amount of tracks.

Most important though, unitrack is a lot more difficult to drive with servos :)

Oh, and the rest of the layout will be peco code 55 as well..

[CaptOblivious]

I missed the compound design in the plan! That's what I get for not zooming in to the image all the way!

[Martijn Meerts]

I'll forgive you this time ;)

I think I'll stick to the initial trackplan. I wanted to put in some new stairs to the attic anyway, and they're not that expensive. Plus, I'll need to replace them anyway if I ever want to set up Tokyo Station, or the 0-scale stuff ;)

[railzilla]

Hi Martijn,

Here a sketch how the tree topology works. This is drawn for Unitrack but the principle can apply to any track. Drawback is with set track you need a lot of different track pieces and with flextrack you need to cut and bend more pieces.  


Modern classify yards of prototype railways are often designed in a similar way.

[Martijn Meerts]

Thanks, I'll try to draw it in Railmodeller and compare it to the current plan to see if there's any difference in length needed. It certainly does look more interesting and prototypical though ;)

[Martijn Meerts]

http://www.jr-chiisai.net/_div/trains_temp/yard_tree_topology.png

So, this is a version using tree topology using the same medium Peco turnouts and track spacing. Tree topology version requires about 20cm less space on each end, meaning I'd save 40cm. Still not that big a deal when you're looking at 510cm total length, but it helps.

I might be able to condense it a little bit more by moving the 6 top tracks further to the right, and making the curve going up to those tracks a bit sharper, bit I think that's only gonna save a centimeter or 2-3.

All in all, I do like the looks of the tree topology better, it's not quite as boring ;)

[railzilla]

Looks nice.
With a rearrangement of the turnouts  you can place tracks 9-14 in such a way that the usable track starts at the same length as track 8. That would save you another 10cm in length on each side. Just by using an other topology you can store additional 4 car commuter on every track.
On the DCC side you could save many servos as on every stage you only need to switch one turnout. First stage has one turnout, second two, third eight, and fourth six. But as the leading turnout is switches it doesn't matter if unused turnouts are being switched too. So by connecting the turnouts on every stage parallel you  could go away with a 4 servo dcc control per fan. A more elaborate solution would be using relays to route the power of the servos. That means seven relays would be needed per fan.

[Martijn Meerts]

I'm not sure if it's possible to drive multiple servo's from the same output on a servo decoder, I'd have to test that. I have an Uhlenbrock servo decoder with 4 servos which I can use to test, considering I probably won't use that for turnouts anyway.

I'm looking at using http://www.esu.eu/produkte/switchpilot/switchpilot-servo/ and http://www.esu.eu/produkte/switchpilot/switchpilot-extension/ The combination of the 2 is great for the Peco turnouts. The question is then, can the switchpilot servo drive multiple servos per output, and can the switchpilot extension handle switching power routing on multiple turnouts per output :)

I don't really want to electronics to get too complicated, and of course things need to be reliable. I've had my share of headaches with standard solenoid driven turnouts in a hidden yard. Many of them just aren't very reliable, or use up too much power (or both ;))
 

As for rearranging, I'm thinking it might be better to move a turnout, so that the top 4 tracks are on 1 fan of 4 tracks, and the 5th and 6th track are on their own. That might save some more space.

[disturbman]

So, this is a version using tree topology using the same medium Peco turnouts and track spacing. Tree topology version requires about 20cm less space on each end, meaning I'd save 40cm. Still not that big a deal when you're looking at 510cm total length, but it helps.

I don't know what you are asking for. 40cm out of 510 it's already something! It fells short of the 10% off but still, it's quite an improvement. A bit too demanding the Martijn? ;)

[Martijn Meerts]

So, this is a version using tree topology using the same medium Peco turnouts and track spacing. Tree topology version requires about 20cm less space on each end, meaning I'd save 40cm. Still not that big a deal when you're looking at 510cm total length, but it helps.

I don't know what you are asking for. 40cm out of 510 it's already something! It fells short of the 10% off but still, it's quite an improvement. A bit too demanding the Martijn? ;)

Well, it needs to fit on the attic one way or another, so every centimeter counts =)

[Martijn Meerts]

It would seem to original, compound design is actually more space saving then a tree topology if you model the entire yard. The length of the turnout block in a tree topology is shorter, but the straight sections start later than on a compound design:

http://www.jr-chiisai.net/_div/trains_temp/yard_compound_vs_tree.png

[railzilla]

You could tweak the tree topology and save the length of one turnout. However both topologies need the same space. If you want to save a lot of space you need to do something more radical and just cut the returning loop. To have enough throughput you could use quadruple track leading to the yard.
A software like Traincontroller should be able to handle the yard. Luckily most Japanese trains are Multiple units, so you do not need to worrry about the position of the locomotive.

[Martijn Meerts]

The problem with cutting the return loop, is that it makes it more difficult to split some of the tracks to allow multiple short trains on 1 track. While still possible, the first train to enter a track will be sitting there for a long time, because it has to wait for the track to fill up with other trains, and then wait some more for the other trains to leave the track. You get a first in last out situation, where you'd optimally want a first in first out situation.

As for the software, I plan to write my own considering I've switched completely to Mac (I gave away my last windows pc a couple of weeks ago), and I don't like any of the current Mac programs available. And while some of my Macs do have Windows installed as well, I really only use it for the occasional game.


Anyways, I went ahead and ordered a bunch of turnouts and some more flex track so I can try some thing with actual track instead of just in railmodeller. Nothing beats trying track configurations with actual track.

[Martijn Meerts]

Well, just when you think you've got it figured out and started buying materials, a new idea comes up. I was looking at the track plan a bit, trying to figure out what length the initial module (or section I guess, since they're not modules that keep to any standard :)), and it struck me. Why not use a combination of a tree topology on 1 half and a compound topology for the other half. A few minutes later I had this plan:

http://www.jr-chiisai.net/_div/trains_temp/yard_tree_compound_combined.jpg

The top 14 tracks are tree topology, and meant for the longer multiple unit trains (16 car shinkansen, M250 and the likes.) The bottom 14 tracks are compound topology, able to store loco hauled trains of various lengths, as well as any other train of various length really.

Track 1 is a through track for trains to get to the compound yard. This track will also get the return loop mechanism. That means I don't need the long loop from the original plan. It does look a bit weird, but it saves me about 0.8 meters in length (4.3 meters vs. 5.1 meters), and I'll need quite a few less turnouts. The tree topology needs to be fine-tuned a little, and of course the track length depends a bit on the actual module lengths.

I could make the compound topology tracks shorter, and thus shorten the entire yard, but I do have some rather long loco hauled trains as well (twilight express, orient express, etc.)

[bill937ca]

Well, just when you think you've got it figured out and started buying materials, a new idea comes up.

That's always the way it is.