Bay 12 Games Forum

Please login or register.

Login with username, password and session length
Advanced search  
Pages: [1] 2

Author Topic: How to build a Multi-cart Ore to Magma Minecart Project without needing power  (Read 16252 times)

WanderingKid

  • Bay Watcher
  • The Overfiend
    • View Profile

So, I've completed, and finally worked out the kinks, for an injury free multi-cart single track minecart system that requires no power whatsoever.  My personal philosophy is that dorfs should only get injured when I tell them to or because they're idiots, not because of fortress design.  I'm relatively proud of the overall design and figured I'd share it with anyone else interested in trying to get a lot of ore down to the magma sea.

Before you ask, yes, I'm familiar that by the time I was done with this thing it's probably about as complex to bring magma up... but that was before I had to fight with the cart return separation system, which took about half the design time of this process while I learned what worked and what didn't.  I've got bits and pieces of this thing all over the forums, too, as I asked for help in different places, thanks to all.  I'm not sure how much of it sunk in, particularly about speed mechanics and corner turn teleportation... but thank you for trying.

So, this system comes in six and a half parts:
1) Staging Area & Track Combination
2) Drop Chute and Catch Rail
3) Dump at magma forges track route and prep to begin imulse elevation
4) Impulse Elevators A, B, and C, necessary because of lack of cavern 'dead space' overlaps.
5) Cart Return System
5a) Cart Holding Loop
6) A touch of dwarven labor

If you're intending to follow this system, I recommend using wooden minecarts to speed up step 6.  Also, make sure to re-designate all tracks as Restricted instead of Low, or it's possible your beards will decide the cart system is a GREAT way to walk down to wherever you're dropping things off at.

Staging Area & Track Combination:
This part is relatively simple, so I'll also describe my routes here.  Each route on the system is a single stop route, the origination.  It is attached to the nearby stockpile and set to haul ores, which the stockpile is also set to, as well as having 3 wheelbarrows.  The hauling stops are set to push east at 75% full, because every now and then one of my bearded dunderheads sticks the wheelbarrow in the minecart along with the ore.  The system cycles fast enough that 4 vs 5 ores in the cart is not a concern.  I've yet to see a full stockpile for any cart except when I stopped the system for tweaking.



First thing I hoped you noticed is the floodgates blocking the rails.  They're attached to the lever below.  When building the cart return system, it can be a bit finicky.  You'll want a way to shut the entire system down with a single lever, then forbid the carts so you don't keep getting push/place orders while you adjust your design once they're in their storage.  It also helps for when you make a slight error and have to shut the system down to retreive an errant cart from the system.

Now, rather rarely you get a bit of a collision on these rails, when two carts try to go at once.  It's ignorable.  The carts are moving slowly enough that they just bounce, a dorf heads over, grabs the stuck cart, sticks it back on the stop, and shoves it again for a retry.  They all feed directly into the drop chute, which is a 3x3 area I was able to find that penetrates all 3 cavern layers without opening me up to cavern critters in the system.  While dropping minecarts onto the heads of flying FBs sounded amusing, maybe next time. 

If you don't have a direct hole to a few levels above your magma forge you can Drop Chute->Catch Ramp->Drop Chute as necessary.  I wouldn't bother trying to dig out a ramp downwards.  Speed goes out of control.  A drop chute drops 9Zs every 5 tics at terminal velocity for a wooden minecart, doesn't spill its contents on landing, and is the fastest way down without building significant infrastructure.  Once a miner starts channeling, though, he's stuck.  Make sure you're accurate with this part.

Drop Chute and Catch Rail
The Drop Chute itself is nothing special.  It's an endless hole, dug by channeling straight down.  The interesting part of the Drop Chute is the catch ramp.

At terminal velocity, the minecart is teleporting over squares completely.  This means when it gets to the ground, a single square ramp may not endow motion to your minecart, as it'll 'teleport' down the ramp, hit the flat track next to it, and stop dead like it was still falling straight down.  Bit wierd, but there ya go.

On a catch ramp, you make sure that one way or another it stops on a ramp, and gets a good head of steam to work with to get to where you want it.



The yellow X here shows you where the drop chute comes down.  The ramp at Z-14 is usually ignored in my system, and it 'catches' at Z-15, where it acts like it had 0 speed and just started midramp.  Z-16 simply gives it a bit of steam, and Z-17 starts my delivery tracks to the magma forges.

Magma Forge Delivery and Impulse Prep
Impulse Elevators don't take a lot of work, per se, to setup, but sometimes you gotta get where you want to start them, and that can be trickier.

The design here is relatively simplistic and won't take long to cover.  The big thing is after the catch ramp you'll need some gentle braking or you're going to be derailing on down-ramps.

First we enter at the top of this diagram, heading east, from the drop chute:


There's two track stops to act as brakes here to keep the system under control, otherwise at the next down ramp after making the turn the system stalls out when it derails, hits the wall, and then drops down onto the ramp to continue.  It's faster to slow the cart down.  The high then the medium give it just enough 'tug' to get the cart back under control.

It drops down to the magma forge layer:

This is about as simple as it gets.  Quantum Storage the ore in a small hole in the wall and pop up the ramp at the bottom back up to Z+1 from the forges.

At Z+1, we've entered the lower portion of this and we're heading north.  Because of the brakes applied earlier, an impulse ramp was needed here to get up the next ramp to get over the drop chute run, so I could start impulse elevator A where I desired.



The next level up is simply a feeder to where I wanted to start the impulse elevator.



Impulse Elevator A, B, and C
My research into impulse elevators is most easily found in another thread where there's a lot of information about what I tried, how I tried it, what worked, and what didn't.  Here I'm simply going to describe how the system works correctly.

The build order is: Flat Turn, Impulse Ramp, Upward Curving Ramp, Upward Straight Ramp, repeat as a mirror.  Phoebeus isn't great for showing directions on ramps, but I'll indicate what's what.  The walls don't have to be constructed, they're leftovers from some failed attempts when I tried to mirror the wiki.  Where you see constructed walls you can just leave the base rock.

At level 0, you begin the process:


On level 1, you have two things.  First, a rocking horse shunt, to get the cart off the rails and stop in case something goes wrong.  Also, a simple upwards ramp to burn off the impulse turn's energy.



Level 2 is a mirror of 0, just reversed:


Level 3 is a mirror of 1


Continue process until reaching desired height.  Once you've gotten there, it will depend on if you're on an even, or an odd, level, which will determine if you've burned off the speed from the impulse turn yet or not.  When you've burned off the speed, you can land it like I did after Impulse A:



The turn is required, otherwise the cart skids to a stop.  The turn re-rails the cart from the skid and it can proceed properly.

If you're on an odd level, like I did for the Impulse Elevator B landing, you have to get a Track Stop involved before your curve to get the cart back under derail speeds, or it will ignore the turn and you may have significant problems.


The elevators are simply long.  Impulse Elevator B is my longest one, 99 Z levels high.  Impulse C is simply a feeder up past the last cavern layer right where I start the drop area.  I'd thought I'd left too much room, but found I'd actually made my life a little easier.  I hadn't realized I'd want a holding pattern loop yet.

Cart Return System
Okay, I'll admit I'm a bit proud of this system.  What it does is let you return x # of carts (5 in this case) onto different tracks dependent on if the return track is already occupied.  It only looks complicated, I promise.

This picture should look familiar from earlier, but it's a bit expanded to show the return rails:


You'll notice that they run into walls.  This is necessary, as you can't build a track stop AND a pressure plate in the same location.  The walls stop the carts just fine, and allow for us to use a pressure plate.  Due to that, this system works.  First, let me show you what it looks like so you can envision what I'm describing:





Those are Impulse Ramps after the hatches to give the carts a boost to get up the ramp to the delivery level.

Ignore the Holding Loop for the moment, I'll explain that in a bit.  First, the cart return system is merely a zig-zag track with a lot of bells and whistles attached to it.  On a longer system, or better timed, bridges could be used here to control track access instead of the system I've designed.  The problem is the 100 tic delay was just too long and I started getting lane stacks which could endanger my haulers, particularly if a lane left a cart down in the splitter system.  I needed a system with more immediate response time.

The next problem is I needed something that would allow access when in the 'off' state, and disallow it when on... such as when a cart was sitting on the pressure plate.  Doors, floodgates... none of these were helpful.  Hatches, however, can cover a track and allow access, and then allow access to a downramp when activated.  They were nearly immediate, but I'd have to design a down-z loop to handle that.  During testing, I found that the high-speed turnaround here would derail the cart at an angle when it came back up to the zig-zag level, making the system useless.  Pushing the down-z loop out an extra square and staggering the turns allowed me to use a wall to correct the diagonal, and the carts were going along happily again.

The final problem is all this zig-zag and rerails really take its toll on a cart, particular with turn friction.  I needed a booster after the second U-turn to keep it from getting stuck between sections on the down-z lane changes.  It had barely enough oomph.  This is one of those places that I'd probably have used roller power for a more controlled boost instead of a ramp, but I was determined at this point to make it work without any power... and it does.

The problem is this system is FAST.  Quite often I'd have carts playing follow the leader onto track one, still causing the exact same issue I avoided using bridges for.  It was quicker, I had less problems, but it didn't remove all the issues.

Enter the holding loop.

Holding Loop
The holding loop works because a hatch opens instantly, but only closes 100 tics (or so) after the plate releases.  So, along the track entering the zig-zag, I built a plate directly behind a trapdoor, which leads to a very simple loop that returns the cart back into the system to try again if it's following one ahead of it too closely.  All my carts 'land' within 100 tics of passing the hatch that controls it's allowance, so a 100 tic separator is plenty for my purposes.  While designing it, I figured out if you can never seem to get a cart out of your holding loop, you've overloaded the # of carts on your system.  Forbid a few until it can clear.

The problem with the holding loop is that if you don't get it quite right, you end up either too fast (and thus jump the down ramps for track selection) or too slow (and jam up the entire damned system).  This is where it gets finicky.  In my particular system, I needed two medium friction track stops to keep carts at roughly the same speed they entered the system in.  That loop could be more contained, I suppose, by adding in zigzags instead of the long running road, but I wasn't sure what I was dealing with and wanted room to work.  I'm sure someone with enough interest will eventually fine tune it to the degree of '14 turns, 32 straights, 1 wall, and a slow track stop' but that's not me.  I can barely understand half the stuff they're talking about when they're trying to describe cart mechanics on tracks at the dwarven physics level.

A little dwarven labor
You'll notice this system is incomplete. No cart actually returns to its start.  That's where your dorfs come in.  They swing over to the cart return system, grab a cart, bring it to the beginning, and start again.  This takes very little time with wooden carts, and what feels like years with a metal one.  One note about this.  Every now and then your beards will think the cart belongs in a stockpile.  Build a nearby one so they don't go wandering off through your fort with a cart you need nearby.


System Testing Made Easy
Once you've got a single cart able to go through your system you need to be able to test all your ramps for possible speed loss.  There's an easy way to do this.  Get all your carts prepped on their start locations, let them fill with the floodgates down, then forbid them all.  Release the floodgates so the tracks are open.  Unforbid the first cart, let it run through the system, then forbid it when it lands on the pressure plate.  It'll stay there without being picked up.  Repeat for carts 2 through X, and adjust as necessary. 

If a cart slips the tracks or has issues, drop the floodgates.  My biggest issue when a cart ran out of speed was it'd bounce between one down-z lane changer and another one.  To get it to release, unforbid the cart in front of it in line, so the hatch pops back up when the cart is returned to its starting location.  The problem cart will settle down eventually and you'll be able to go in and get it.  The final delivery system is a bit finicky, as I mentioned, and will take a little adjustment depending on your particular final design.



If you found this helpful, if you have comments, or just generally want to tell me I wasted a few hours putting this together, let me know.  Thanks for reading through this.

itg

  • Bay Watcher
    • View Profile

Nice work! Very well-written guide, too. I'm curious, have you tried stacking minecarts for a cart return system? I've noticed in my own little experiments that if a minecart falls on another, it floats on top of it. If you drop in more minecarts, all but the original cart pile up on the second level, "quantum stockpile" style. If you send in a dwarf to carry off the bottom cart, just one floating cart falls down to fill the vacuum, apparently without chance of falling on the dwarf. It seems like it could make for an safe, unlimited capacity return system, but I haven't had an occasion to test it.

Ravendarksky

  • Bay Watcher
    • View Profile
    • DFMon.exe - Get rid of DF SPAM

Thank you for this, my dream of an automated mine cart obsidian caster can now be realized using your multi cart return system (to deliver the obsidian to different squares)

I have a similar system to yours except that I'm bringing magma to the surface (well 10 z levels in the sky) instead of bringing ore down.

I know you aren't using any power, but I found the best way to stop carts and have them travel quickly was to drop them straight down vertically and land on a roller.
« Last Edit: August 09, 2013, 07:29:23 am by Ravendarksky »
Logged

Maw

  • Bay Watcher
    • View Profile

<deep voice>

Impressive.

Most Impressive.
Logged
The three stages of information assimilation in bay 12:
1)horror
2)curiosity
3)weaponization

edgefigaro

  • Bay Watcher
    • View Profile

<deep voice>

Impressive.

Most Impressive.
Yeah. This is some quality engineering.
Logged

Di

  • Bay Watcher
    • View Profile

This is a masterful dwarven elevator.
All engineering is of the highest quality.
Well done.

Though I'd try building pressure plates linked to track-selection hatch all along the path instead of holding loop. So that the firs cart would send the next ones immediately onto next path instead of loop.
Logged
Quote from: Creamcorn
Dwarf Fortress: Where you meet the limit of your imagination, moral compass, sanity and CPU processor.
http://www.bay12forums.com/smf/index.php?topic=103080.0 Fix sober vampires!
http://www.bay12forums.com/smf/index.php?topic=91442.0 Dwarven Cognitive Science

Larix

  • Bay Watcher
    • View Profile

Good stuff. I'm astonished that falling minecarts can actually be caught in such a way that they keep hold of their contents, nicely engineered there.

I also find it interesting that you came up with similar features for controlling up-moving minecarts that i ended up using as well, like the corner directly behind the up-ramp, and encountered (and conquered) problems i'm all too familiar with like unexpected derailing and diagonal movement.

I particularly like the 'go to first open station' switch, that's an elegant use of the possibilities minecart processing offers, without branching off into alternative logics. It's a bit of a shame dwarfs will only ever accept one cart per route, so returning carts directly to their loading stations would have required weight-sensitive switches and different-weight carts. The switch logic wouldn't have been much different, though.

I know i'll just stick with straight downward ramps and the plain two-wide ramp elevator. They take little engineering, just the occasional speed limiter when going down and an eye for viable paths. But it's nice to see what can be done with a bit of dedication.

@Di: the problem is that a pressure plate along the track only sends a 'pulse', an open signal followed by a close 100 steps later. An immediately following cart will get switched to another station, one that follows after 100+ steps will not get switched. For a properly 'held' switch you need a pressure plate on which the cart _stands_.
Logged

WanderingKid

  • Bay Watcher
  • The Overfiend
    • View Profile

Nice work! Very well-written guide, too. I'm curious, have you tried stacking minecarts for a cart return system? I've noticed in my own little experiments that if a minecart falls on another, it floats on top of it. If you drop in more minecarts, all but the original cart pile up on the second level, "quantum stockpile" style.

According to the minecart wiki, it's been noted that carts can get 'stuck' when quantum stockpiled this way.  I wanted to avoid this so I could have a system that could be left completely unattended.
Thank you for this, my dream of an automated mine cart obsidian caster can now be realized using your multi cart return system (to deliver the obsidian to different squares)

That sounds quite interesting.  Hopefully you'll share some ideas from that inspiration. :)

Though I'd try building pressure plates linked to track-selection hatch all along the path instead of holding loop. So that the firs cart would send the next ones immediately onto next path instead of loop.

The problem with this is you can't trust dwarven labor.  They get to the cart when they get to it.  Under normal load, only the first two rails on the return system are usually used, but I've seen 4 carts in the return quite often and occassionally 5.  The dorfs reset them pretty quickly, but they also load fast.  If you have the plate on the path, you only have 100 tics worth of safety on that particular rail return, and that just isn't enough.  You need the particular return line permanently unavailable until the dorf gets the cart clear, or you end up with activation issues.

AutomataKittay

  • Bay Watcher
  • Grinding gears
    • View Profile

I'm impressed with cart return system, it's pretty dwarf-proof at their speed :D
Logged

WanderingKid

  • Bay Watcher
  • The Overfiend
    • View Profile

One thing I should probably mention.  In my fort, the magma forges are roughly ~150 steps from the ore layer by main stairwell.  In 4 years I moved about 10 pages of ore (seen via (k)) to a quantum stockpile down there before I realized I needed this system.  This system has matched it in roughly 1 year, and that was shutdown for approximately half of the year for tweaking the cart returns at varying points.  I can't give you exact numbers but it has sped ore delivery by an exponential factor, not simply a multiplier, so it was definately time well spent.

I've got a LOT of ore.

Good stuff. I'm astonished that falling minecarts can actually be caught in such a way that they keep hold of their contents, nicely engineered there.
Thank you.  I was surprised too so my first tests were centered around filling carts with various things and dropping them down the hole to see what fell out.  So far, nothing has.

Quote
I particularly like the 'go to first open station' switch, that's an elegant use of the possibilities minecart processing offers, without branching off into alternative logics. It's a bit of a shame dwarfs will only ever accept one cart per route, so returning carts directly to their loading stations would have required weight-sensitive switches and different-weight carts. The switch logic wouldn't have been much different, though.
Yeah, I debated on going with the weight processing, and would merely require using an 'allowed' switch in front of a drop hatch that ran directly to the landing and keeping the zig-zag on a single Z.  I decided to KISS the system first, then perhaps come back to that idea in another fort.

Scarlet_Avenger

  • Bay Watcher
  • Cancels drink: Too insane
    • View Profile

Would it be too much to ask for a download?

I need to see use this!
Logged
Quote
Well given that we started building computers inside of DF years ago, it's really only been a matter of time until the dwarves figured out how to utilize them for their own purposes.

Skynet has nothing on DF 40.01. Frankly, I'd be less horrified at living with John Connor than anyone named Urist.
\_(ツ)_/

WanderingKid

  • Bay Watcher
  • The Overfiend
    • View Profile

A note from further testing for full disclosure:

There is something off in the timing in this system.  I'm not entirely sure what's causing it but if four carts go into the system too close to each other, the last cart ends up practically flying out of the final elevator.  I find myself having to 'drop in' on the system occassionally to double check that nothing's gone wrong.

It happens maybe once a dwarven year, but it's worth noting for full disclosure.  My guess is it's something to do with the drop chute ending up with a quantum stack, and a falling cart bumping a stuck one and it going to higher speeds than the system expects, but I haven't been able to catch it in the act when the over-speeding starts.  It could be on one of the breaking areas too, where a cart is being bumped over the brakes.  I have yet to have a cart stuck IN the system, just in the minecart delivery area, but it's worth noting.

Because of this I'd be wary of using more than 5 carts in any multi-cart system that includes a drop chute.  At 6 carts I nearly trebled the occurences of carts lost in the cart return system because of this issue.

Karakzon

  • Bay Watcher
  • [ethics:give a shit?: denied]
    • View Profile
Re: How to build a Multi-cart Ore to Magma Minecart Project without needing power
« Reply #13 on: September 10, 2013, 09:42:02 am »

Just a note on dwarves pathing into the inner workings, wouldn't it be possible to prevent it by including smaller 1 or two z seclusion drops in one or two places so they cannot form a path into the cut off areas? Then you can just limit access with a door to specific sections and have nothing but any flyer's try to path up through it, instead of having to restrict everything.

Would make the process take slightly longer, but it might be safer for those who are lazy with restrictions.
Logged
I am Dyslexic. No its not going to change any time soon.
Bolts of Exsanguination THE terrifying glacier export, get yours today!

WanderingKid

  • Bay Watcher
  • The Overfiend
    • View Profile
Re: How to build a Multi-cart Ore to Magma Minecart Project without needing power
« Reply #14 on: September 10, 2013, 01:24:19 pm »

Absolutely.  As a pathing deterrent, that would definately work during the return system.
Pages: [1] 2