just a visual interface, where you can make all kinds of logic combinations
one box that receives all the input and sends the results out to the wire network
just a visual interface, where you can make all kinds of logic combinations
I don't think you quite understood what I said, I can clarify a bit.Honktown wrote: βThu Oct 10, 2019 8:24 amA latching mechanism connected to a power switch that goes turns off if there's a fuel cell signal (read contents, pulse read) and turns on when there's something else. A basic SR latch would do it (fuel cell = reset), where instead of using the same trigger signal like high/low value of an accumulator you use two different ones. I'm assuming an inserter only generates a pulsed signal when it grabs something, and not if it turns on with something in its grabber.Cribbit wrote: βThu Oct 10, 2019 5:21 amI really wanted to make a 2-gap (2 tiles from reactor to first heat exchanger) fuel loading that relied on inserters that shut themselves off as soon as they grabbed fuel, and then finished their swing on another signal. With a 3 gap or with chests instead of power off you can pretty easily get it working.
I only used long arms in my reactors because I figured it was more efficient/reliable for heat pipes to run directly around the reactor.
Thinking now, why are you trying to disable the arm mid swing? You never said why, and that really makes it pointless why you want this. Also, however you try and design a super-optimized perfectly done thing, you've gotta get that fuel in/out somehow, so there will be a lump of a roboport and annoyance somewhere else.
An interesting and horrible idea: use stack inserters to put down fuel, and use basic inserters + fast belt so they don't have time to grab more than one fuel cell. Worse, if for some reason you have like 50 nuke reactors, pulse-width-modulate power to the inserters so they don't have time to grab more than one cell.
Honestly it'd be nice if there was a hysteresis/SR combinator, since that's probably what 90% of the practical use cases for the circuit elements are.
I was thinking today if the conditions for combinators showed in the detail pane. I don't blueprint much so it's easy to mess up a condition. They might show something if an option is enabled for logistic view, didn't try it yetBlueTemplar wrote: βWed Oct 09, 2019 8:34 amSpeaking of which, it would be nice if there was some visual feedback when hovering a wire over a combinator or power switch that would show whether it's going to connect to input or output - would allow for even faster, error-free connections !
The point is to only have a 2-gap. Of course it's an easier problem with a bigger gap.Honktown wrote: βThu Oct 17, 2019 3:05 amOh, I always used requester chests for the cells and either passive providers or active providers for the used cells.
Chests can be linked perpendicularly to the reactors by inserters (i.e. in the same direction the line of reactors is going), and heat pipes can run under long arms, so they fuel can still be a minimum distance away.
I though of a way to cut power mid-swing: use two power switches in an OR set-up. If the hand reads fuel cell (each arm), disable one power switch, and if the steam is low (all arms), enable the other. How close are you putting switches? I usually use a big power pole or a bunch of medium ones and put them further away. Utility > compactness. Also, you can isolate the power poles, and connect them all to one switch, ya know? I did that a lot because I hated setting each arm individually (if you read steam directly you have to change each one when you increase power... ugh). If synchronized properly, you could read belts and when there's exactly the right count, enable the arms. That'd be awful to figure out.
This is all just an exercsise in blueprint showing-off right? Only with uranium set to super low or marathon would I see problems throwing down more miners or researching kovarex.
I was thinking today if the conditions for combinators showed in the detail pane. I don't blueprint much so it's easy to mess up a condition. They might show something if an option is enabled for logistic view, didn't try it yetBlueTemplar wrote: βWed Oct 09, 2019 8:34 amSpeaking of which, it would be nice if there was some visual feedback when hovering a wire over a combinator or power switch that would show whether it's going to connect to input or output - would allow for even faster, error-free connections !
You say 2-gap and a power switch can't fit, how does that make any sense? All I can find when I look up 2-gap is a lane balancer. Be clearer, please.Cribbit wrote: βThu Oct 17, 2019 3:25 amThe point is to only have a 2-gap. Of course it's an easier problem with a bigger gap.
This only works with 1 power switch per reactor. You can't do that on a 2 gap, because switches are 2x2.
You can't have roboports, because they're bigger than 2x2.
You can't use long arm because it can't filter when it needs to remove dead cells.
Luckily, belts don't actually hold that many fuel cells in the "on belt buffer", so I just went with the more straightforward dead cell latch.
Point still stands - 2x2 power switches makes them hard to fit into anything interesting.
I explained 2-gap earlier. In the context of nuclear reactors it's a pretty straightforward concept - 2 tiles from reactor to the start of the first heat exchanger. You need this in order to have decently compact reactors without using a ton of pumps because steam starts to overrun pipe throughput.Honktown wrote: βThu Oct 17, 2019 12:02 pmYou say 2-gap and a power switch can't fit, how does that make any sense? All I can find when I look up 2-gap is a lane balancer. Be clearer, please.Cribbit wrote: βThu Oct 17, 2019 3:25 amThe point is to only have a 2-gap. Of course it's an easier problem with a bigger gap.
This only works with 1 power switch per reactor. You can't do that on a 2 gap, because switches are 2x2.
You can't have roboports, because they're bigger than 2x2.
You can't use long arm because it can't filter when it needs to remove dead cells.
Luckily, belts don't actually hold that many fuel cells in the "on belt buffer", so I just went with the more straightforward dead cell latch.
Point still stands - 2x2 power switches makes them hard to fit into anything interesting.
There's absolutely no need to know when to remove a dead cell, it's trash, it should always remove it. The other option is remove it when you go to insert a new cell. It's always 1:1 inserted cells to dead cells, and it can be removed at the same time a new cell is inserted without desyncing.
An important point is reactors transfer heat through each-other, so it's not like you need to have every reactor connected. (a single heat pipe might only transfer like 160MW over short distances, which is a concern)
I forgot last night, some (all?) things lose their circuit signal when they have no power, so if you read when an inserter had a cell in it's hand, you'd have to latch it somewhere else.
You should ask the devs to add a power meter to the circuit network. That way we can finally get rid of all this steam tank measuring business and insert cells when we've accumulated enough gigajoules to warrant turning on the reactor again. Not like power meters are space-age sci-fi technology.
Split off from the belt into a dead-end yellow belt. That will act as local buffer and can be read for fuel contents. A chest can also be fed from a belt for this purpose, but the additional inserters required make it less compact.Cribbit wrote: βWed Oct 16, 2019 6:07 pmThe "why" is fuel loading, namely the standard "steam buffer" nuclear where you only load more fuel (into every reactor at once) when you get low on steam. Generally, a belt rings the entire set of reactors. The easy method is to simply fill one side of the belt entirely with fuel and then have some sort of mechanism to insert when needed. However, if you don't have enough fuel to fill half of a belt you quickly run into issues where activating the inserters means there might be a gap in fuel, resulting in that reactor not activating.
That can't be done on a 2-gap, inherently.foamy wrote: βThu Oct 17, 2019 5:10 pmSplit off from the belt into a dead-end yellow belt. That will act as local buffer and can be read for fuel contents. A chest can also be fed from a belt for this purpose, but the additional inserters required make it less compact.Cribbit wrote: βWed Oct 16, 2019 6:07 pmThe "why" is fuel loading, namely the standard "steam buffer" nuclear where you only load more fuel (into every reactor at once) when you get low on steam. Generally, a belt rings the entire set of reactors. The easy method is to simply fill one side of the belt entirely with fuel and then have some sort of mechanism to insert when needed. However, if you don't have enough fuel to fill half of a belt you quickly run into issues where activating the inserters means there might be a gap in fuel, resulting in that reactor not activating.
This, incidentally, is my setup:
Not as yet. The idea of putting up fifteen thousand or so radars is kind of daunting.
Plumbing isn't related to what the reactor distance is. Only heat transfer through the heat pipes is. As long as you're getting heat to the exchangers, the exchangers could be 2 tiles away, they could be 30 tiles away.Cribbit wrote: βThu Oct 17, 2019 1:22 pmI explained 2-gap earlier. In the context of nuclear reactors it's a pretty straightforward concept - 2 tiles from reactor to the start of the first heat exchanger. You need this in order to have decently compact reactors without using a ton of pumps because steam starts to overrun pipe throughput.
Not sure where the dead cell comment comes from. The long inserter chest thing? If you are using a chain of long inserters and chests, then you physically cannot remove the dead cells. Or am I missing something there? My "filter" comment was that even if you figure out a way to use the same line of chests the long inserters have no way to differentiate moving a dead cell and a new cell.
By nature of heat pipe distance you end up needing to connect every 6 tiles, just under once per reactor. You have very little flexibility on where the heat pipes physically go.
Power meter still wouldn't help for this, since there's a significant lag between when fuel is put in and when steam actually starts being produced. The only thing that would make this simpler is if the circuit network could control reactors directly.
<image>
This is my current nuclear setup. Please, show me how you could fit power switches in there, under any setup.
You're right, plumbing only matters for how many heat exchangers are in a row. My mistake.Honktown wrote: βThu Oct 17, 2019 8:27 pmPlumbing isn't related to what the reactor distance is. Only heat transfer through the heat pipes is. As long as you're getting heat to the exchangers, the exchangers could be 2 tiles away, they could be 30 tiles away.Cribbit wrote: βThu Oct 17, 2019 1:22 pmI explained 2-gap earlier. In the context of nuclear reactors it's a pretty straightforward concept - 2 tiles from reactor to the start of the first heat exchanger. You need this in order to have decently compact reactors without using a ton of pumps because steam starts to overrun pipe throughput.
Not sure where the dead cell comment comes from. The long inserter chest thing? If you are using a chain of long inserters and chests, then you physically cannot remove the dead cells. Or am I missing something there? My "filter" comment was that even if you figure out a way to use the same line of chests the long inserters have no way to differentiate moving a dead cell and a new cell.
By nature of heat pipe distance you end up needing to connect every 6 tiles, just under once per reactor. You have very little flexibility on where the heat pipes physically go.
Power meter still wouldn't help for this, since there's a significant lag between when fuel is put in and when steam actually starts being produced. The only thing that would make this simpler is if the circuit network could control reactors directly.
<image>
This is my current nuclear setup. Please, show me how you could fit power switches in there, under any setup.
The dead cells bit is because it sounded like you're basing part of the circuit on when you're removing cells ("dead cell latch"), and said you don't know when to remove dead cells. There's no point in knowing when.
Playing with inserters and chests, turns out it won't work because you have to insert into chests (I was using chests because it guarantees there's a cell in front of the arm). I did learn something very interesting, inserters can insert onto and take from a space with a heat pipe.
A power meter would work, because unless you're accounting for literally everything, including how many heat pipes there are, variance in energy consumption, how fast heat's going down the lane, how quickly you're consuming power during the warm-up, etc, there will be discrepancies in energy across the system. Cells provide exactly a certain amount of energy. The tanks would have to be overbuilt, but you can always keep exactly the power required and never waste any.
In your image, you have extra heat pipes (the 3 merging to 1). It'd be assymetrical/off-ratio (oh god the OCD/aspergers) but you don't need that many. Only one connection is required per reactor, even with full neighbor bonus. A fully neighbor bonus'ed reactor would generate 160 megawatts, delivering to 2x8 heat exchangers, which is easily what a heat pipe can handle (the pump ratio would be off if that's bothersome, cause you'd need 1 1/3rd pumps minimum).
A quick test, medium power poles can connect a circuit wire 12 tiles distant to an element, and big power poles 9. Didn't expect that.
Power meter + integrator (or one that is both). Once you accumulated x gigajoules (x being how much power you generated from the connected reactors), insert another fuel cell and reset the integrator. Almost the same as using the steam tanks, but much easier on the calculations. If you whipped up a good spreadsheet, you could have number of tanks, steam needed for minimum power at max drain, a proper window so you don't repeatedly insert two cells or fail to insert a new cell, etc. Tanks don't fill evenly without pumps which is super annoying, but the calculations could be a little over for security.Cribbit wrote: βThu Oct 17, 2019 9:18 pmYou're right, plumbing only matters for how many heat exchangers are in a row. My mistake.
My point about dead cells is the same thing you just realized - if you use long arm inserters to chain chests, you will have a very hard time removing dead cells. I'm not entirely sure it's even possible, you'd definitely have to mix in belts and certainly wouldn't work in a 2 gap.
A power meter would simply be a yes/no is enough power being produced right now at this moment. For nuclear, you only dip below that threshold when completely out of steam. You then have 0 steam to generate power with until enough heat spreads to generate more steam. Hence, a power meter would not help for nuclear, since you have to measure steam buffer, not current power output.
I know I have more than necessary on the 3:1 pipes, but even without that there's still not enough room. They're only in the design for aesthetic purposes, as you note.
Cribbit wrote: βThu Oct 17, 2019 9:18 pmYou're right, plumbing only matters for how many heat exchangers are in a row. My mistake.
My point about dead cells is the same thing you just realized - if you use long arm inserters to chain chests, you will have a very hard time removing dead cells. I'm not entirely sure it's even possible, you'd definitely have to mix in belts and certainly wouldn't work in a 2 gap.
A power meter would simply be a yes/no is enough power being produced right now at this moment. For nuclear, you only dip below that threshold when completely out of steam. You then have 0 steam to generate power with until enough heat spreads to generate more steam. Hence, a power meter would not help for nuclear, since you have to measure steam buffer, not current power output.
I know I have more than necessary on the 3:1 pipes, but even without that there's still not enough room. They're only in the design for aesthetic purposes, as you note.
Not that guy, but I never thought about a brownout alarm and then switch to back-up steam. Normally one compares steam engines and tanks vs accumulators for early storage, but steam turbines and tanks would be exceedingly efficient for late storage (2.425 GJ and 5.82 MW deliverable). Never ran out of fuel before, but I have had issues with exceeding power requirements or doing something ADD and forgetting to finish a circuit or whatever. Would also be useful in a resource-tight game, one could "overbuild" turbines and tanks for the accumulator phase, and then use alarms when the steam production isn't enough to keep up with spikes of drain. Then hope you have enough boilers finished by then, or start switching crap off.foamy wrote: βThu Oct 17, 2019 11:31 pmNah, you can totally brownout a nuclear plant if you have a large enough base.
I settled on a dual-phase steam tank setup (which also gave me the room to stuff in a roboport & accumulator bank to self-power). The first set of tanks are the measuring ones, which are emptied by pumps into a second buffer tank, from which the turbines drain. This guarantees that the measuring tanks are always drawn from first. The measuring tanks will also only start declining if draw from the turbines exceeds the production of the exchangers, which means that some of the exchangers (though not necessarily all: it depends on the load) have stopped producing steam. If the contents of the measuring tanks are below 90%, a signal is generated which is ANDed with a timer signal, triggering the loading inserters if, and only if, both 200 seconds or more have passed since the last time the inserters were triggered and the steam level has started dropping. The two steam tanks can then, if needed, supply the turbines for ~20 seconds with absolutely no further replenishment while heat propagates back along the pipes and re-enables the exchangers. In practice the exchangers start producing steam again nearly immediately.
I have separate brownout-detection circuitry designed to cut the plant's internal operations (pumps & inserters) off the main grid and run them on a dedicated set of turbines & accumulators until such time as the main grid's own accumulators have exceeded the level of the plant's internal ones.
It isn't back up steam precisely -- I have a separate plant for that, which I believe I described to you in a different thread -- but the nuclear plant was designed to self-power itself and guarantee it works under all loads, even if the rest of the factory is in brownout conditions.Honktown wrote: βFri Oct 18, 2019 12:11 am
Not that guy, but I never thought about a brownout alarm and then switch to back-up steam. Normally one compares steam engines and tanks vs accumulators for early storage, but steam turbines and tanks would be exceedingly efficient for late storage (2.425 GJ and 5.82 MW deliverable). Never ran out of fuel before, but I have had issues with exceeding power requirements or doing something ADD and forgetting to finish a circuit or whatever. Would also be useful in a resource-tight game, one could "overbuild" turbines and tanks for the accumulator phase, and then use alarms when the steam production isn't enough to keep up with spikes of drain. Then hope you have enough boilers finished by then, or start switching crap off.
One could also prepare a reactor/expansion that way, and add the reactors and boilers later. I don't blueprint so that might be a good plan. Thanks!
IIRC heat pipes are actually the densest form of energy storage, although that probably declines if you have to have a separate set of heat exchangers for them rather than putting the heat back into your main network (which you would have to, by lack of circuit network heat switch).Honktown wrote: βFri Oct 18, 2019 12:11 am
Not that guy, but I never thought about a brownout alarm and then switch to back-up steam. Normally one compares steam engines and tanks vs accumulators for early storage, but steam turbines and tanks would be exceedingly efficient for late storage (2.425 GJ and 5.82 MW deliverable). Never ran out of fuel before, but I have had issues with exceeding power requirements or doing something ADD and forgetting to finish a circuit or whatever. Would also be useful in a resource-tight game, one could "overbuild" turbines and tanks for the accumulator phase, and then use alarms when the steam production isn't enough to keep up with spikes of drain. Then hope you have enough boilers finished by then, or start switching crap off.
One could also prepare a reactor/expansion that way, and add the reactors and boilers later. I don't blueprint so that might be a good plan. Thanks!
I've considered using solar for nuke plants, completely isolated. Pumps/inserters/etc don't take much, and even roboports have a "drain" and it's not near the cost of active bots. I've used a belt before to feed repair kits from a main-base logistic network over to a reactor logistic network (kept bots from traveling 20 miles to repair things). Same could be done for uranium cells and used uranium cells. Would also give a source of accumulators, since turbines take priority and if the accumulators ever got hit you'd know the nuke plant wasn't enough and can isolate the plant with a latch.
Ah, true about back-feeding the steam. I've thought about using pumps on the "input"/"output" of a separate storage bank before, just not in this context. Would need to watch for throughput issues which makes things much more complicated (separate pump for every 20 turbines/pipe paths and by golly do you need a lot of those turbines in vanilla).Cribbit wrote: βFri Oct 18, 2019 12:38 amIIRC heat pipes are actually the densest form of energy storage, although that probably declines if you have to have a separate set of heat exchangers for them rather than putting the heat back into your main network (which you would have to, by lack of circuit network heat switch).Honktown wrote: βFri Oct 18, 2019 12:11 am
Not that guy, but I never thought about a brownout alarm and then switch to back-up steam. Normally one compares steam engines and tanks vs accumulators for early storage, but steam turbines and tanks would be exceedingly efficient for late storage (2.425 GJ and 5.82 MW deliverable). Never ran out of fuel before, but I have had issues with exceeding power requirements or doing something ADD and forgetting to finish a circuit or whatever. Would also be useful in a resource-tight game, one could "overbuild" turbines and tanks for the accumulator phase, and then use alarms when the steam production isn't enough to keep up with spikes of drain. Then hope you have enough boilers finished by then, or start switching crap off.
One could also prepare a reactor/expansion that way, and add the reactors and boilers later. I don't blueprint so that might be a good plan. Thanks!
If you set it up right you wouldn't need to duplicate turbines, you would just pump the emergency steam into your main network. Could easily even have the emergency storage refill once fuel was restored, so you'd only have to fix the reactors, not actually muck with the backup.
Don't need the solar panels, since accumulators are last in any case, and this way the design's compatible with a no-panel run. I could fire you the blueprints so's you can see it in operation, if you like.Honktown wrote: βFri Oct 18, 2019 2:04 amI've considered using solar for nuke plants, completely isolated. Pumps/inserters/etc don't take much, and even roboports have a "drain" and it's not near the cost of active bots. I've used a belt before to feed repair kits from a main-base logistic network over to a reactor logistic network (kept bots from traveling 20 miles to repair things). Same could be done for uranium cells and used uranium cells. Would also give a source of accumulators, since turbines take priority and if the accumulators ever got hit you'd know the nuke plant wasn't enough and can isolate the plant with a latch.
Yes, the per-inserter power switch is what kills it for a 2-gap.Honktown wrote: βFri Oct 18, 2019 2:04 amDon't want to pull off-topic from OP. A big complication with what's wanted here is not only do you need to watch a signal, you need to know which arm is active. That's an individual latch+power switch for each arm if switching power (two dimensions of information). Belt readers might be useful. If all belts read a fuel-cell, you can insert (one or two, not sure about tick progress/inserter reach). More accurately, the last belt could stop if it has a fuel cell, the next would need to know the previous stopped, down the line. This might not be cascadeable.
My original thought was a completely isolated reactor, since their energy cost is relatively fixed over their lifetime and low in itself, even with pumps all over the place. (technically if you're regulating power they could go up a little bit while consuming more often, but it's not a lot).
A problem with belt-buffering is the odd spacing/timing. I'm assuming an ultimate goal here is a blueprintable reactor, so both sides are effectively separated (maybe not). There's also the issue of time to get cells in front of reactors, which grows worse if more reactors are chained. A way around it is to increase the minimum steam value, to give a longer consumption time before cells are inserted. With belt spacing, you'd have to insert at a certain interval related to belt speed. This is all turbo-optimizing, because a fuel cell on a belt is storage, it doesn't waste energy or resources, minus the cost on creation and productivity (I've accidentally filled a steel chest with blue splitters, that hurts but I obviously didn't notice at the time).Cribbit wrote: βFri Oct 18, 2019 2:44 amYes, the per-inserter power switch is what kills it for a 2-gap.
Doing it at the belt level to space out fuel is actually really interesting and totally feasible.
You simply have each section of belt turn off if the next section has a fuel cell in it. The last belt section turns off if itself has a fuel cell in it. Assuming you're doing fuel inner edge, dead cells outer edge, all on one belt, the dead cells will still clear out easily since the belts have to cycle to move the next fuel cell along. This simple setup means that there would only be 1 fuel cell per belt rather than 8, an 8x reduction in "on belt buffer".
I'm sure even further improvements - probably even 1 cell per reactor, stopped on the section infront of its inserter - is feasible, I'm just way too tired to figure it out at the moment. Probably similar wiring, just wiring the section infront of a reactor to all the sections until the next reactor. Undergrounds kind of mess it up but still.
Power switching on inserters looks cooler though.