Posts by Gus_Smedstad

    It’s true that I’ve seen things in snippets of code that raise my hackles. I’m thinking specifically of the bit that checks if you’re right-clicking on a machine with a bucket of water, and removes any paint if you are.


    Said code creates a new water bucket object on the heap every time you right-click on a machine, and compares the object to what you’re holding. While I don’t know the details of the API, there has to be a way to do that with a simple comparison against a constant. I.e. get the type ID, or call a method with the type ID as a parameter.

    I suspected the pipes as well, given that every pipe segment is constantly updating, and I’ve got an awful lot of pipes. The hand scanner claimed they weren’t causing delays, but I don’t know if I really trust the scanner on this issue.


    Minecraft’s problem, and the problem with Minecraft mods, is Java. Java’s bytecode is just never going to be as fast as programs compiled to actual machine code. So-called “just in time” compilation can help, but there’s still the overhead of doing the compilation at run time rather than at development time, and that’s never going away.

    I think I’m taking a break from Gregtech.


    I starting thinking about this because the only projects on my horizon are building an assembly line so I can make a quantum suit, making a high octane gasoline synthesis line, improving my oil refinery to meet the needs of the gasoline production, and making a large combustion engine to burn it.


    The quantum suit would be primarily to go after the Ender Dragon. That would give me access to naquadah, but I’m not feeling the urgency of that.


    The assembly line requires 4K EU/tick, which means upgrading at least some of my energy storage and distribution to 8kV tech. Which in turn means a lot of waiting for tungsten to smelt.


    Given the output of the thorium reactor, the gasoline stuff would just be for the experience of building it. Right now the reactor is doing a decent job of covering all my energy needs.


    Another issue is that I’m experiencing pretty severe frame rate issues. I’ve got a 8700K CPU and a GTX 1070ti GPU, but I’m seeing frame rates dip to 6-7 FPS. Turning down some of the graphic settings got that up to 10-11 FPS, but it’s a real playability issue. Interactions with chests are very slow when the frame rate drops below 9, and the light flickers.


    I don’t know how to fix it. It’s worse when I’m looking in some directions, but it’s true even if there’s a wall directly in front of me, so it’s not a simple graphical update issue. I’m assuming it’s some entities triggering lots of block updates.


    I’m not sure when, or if, I’ll return.

    The Large Heat Exchanger / Turbine setup does not appear to destroy water. The large boiler / turbine setup does, so clearly the problem is purely large boilers. I don’t think it’s a startup problem with the boilers, I think they just consume more water per L of steam than the turbines return.


    Regarding the LHE / turbine steam supply / demand rate matching, my primary concern is water destruction, rather than loss of power. Though one thing about the steam buffer is that the excess steam eventually converts to power. When the battery gets close to full the system shuts the reactor off, but the steam in the buffer continues to run the turbines for a while, generating power. While the system will also shut off if the steam buffer gets full, in practice I’ve yet to see it get more than about 10% full before the battery is full.


    Of course, if I were to flip the “ignore battery state” lever, eventually it would happen.


    One way-shutters in Gregtech valves are, generally speaking, as effective as pumps in increasing fluid throughput. At least, for a steady flow. Pumps are more effective at the start and end of a cycle. At the start, the flow in a shuttered pipe sequence starts at 1/2 the input rate and builds toward the full rate; at the end, it falls in an inverse geometric rate, so each second the rate is 1/2 what it was in the prior second. Pumps deliver the full rate at both start and end, provided can handle that rate. Shutters, of course, have no rate limit.

    Main train station, showing added signals, and two trains. The steam locomotive (yellow, background) is waiting for the electric locomotive (red, foreground), which I've locked in place for this image.



    Above ground ore depot for lithium. Note water collection for manufacturing of drilling fluid on site.



    Oil refinery - gives you some idea why I'm feeling like GT5U is a plumbing nightmare. Brown = Heavy fuel or Toluene, Yellow = Light Fuel, Red = Naphtha or Hydrogen, Orange = Sulfuric Acid of Hydrogen Sulfide, Green = Oxygen, Brown / white = Creosote oil.




    Spray paint room. 16 tanks, one for each chemical dye, one chemical reactor to make dye, one fluid canner to fill spray cans. Since dye doesn't render correctly, there are painted wool sections above each tank to show color.



    I’m not really that thrilled with the inclusion of a steam buffer in my reactor setup. I just don’t see a better solution.


    The way I see it, the problems are:


    1. Reactor heat output is somewhat irregular. My thorium reactor produces about 950 HU/sec, but it’s not a steady figure. Eyeballing it, it appears to be +/- 30 HU/sec.


    2. You can’t safely throttle the hot coolant output of a fluid reactor. If you regulate output to anything less than the actual average, eventually the hot coolant tank fills up and the reactor starts heating up. This is fixable with Nuclear Control - I have a temperature safety on my reactor even though this should never happen - but you’re throwing away heat if you do this.


    One possible solution is an external hot coolant buffer. You can test if that’s nearly full, and shut down the reactor before the internal tank starts backing up.


    3. Steam production is unpredictable because of (1) and (2).


    4. Because turbines lose efficiency if starved, it’s best to have a slight overproduction of steam relative to turbine consumption, rather than a deficit. Matching it exactly isn’t really possible.


    5. You can tune fluid regulators on the turbines to slightly overconsume steam, but that’s almost certainly going to irregularly starve some turbines. The more turbines in your setup, the more likely that one turbine is going to see a significant deficit.


    For example, my reactor appears to produce 76,000 L/sec of steam, +/- 2,400 L. Turbines 1 and 2 consume 24,000 L each, turbine 3 consumes 27,000, for a total of 75,000. If I tune for the maximum (78,400) by adding 1000 L consumption to each, turbine 3 might only get 23,600 L, a shortfall of 3,400 L.


    6. A buffer solves this problem, and lets you run all turbines at exactly 100%. You can easily see if you’re over or under producing steam over time, smoothing over small production irregularities (to a limit, obviously it didn’t solve the huge swings I was seeing).


    7. If the system is tuned to slightly overproduce, eventually you want to shut off for a short period to avoid steam overflow (and destruction). You can’t really test this without a fair sized tank, flow in pipes is too irregular.


    8. Most tank mods are pretty low capacity. Railcraft’s steel tanks hold more than almost anything else, but have a different drawback - a limit of 20,000 L/sec per valve. My setup requires a minimum of 4 input and 4 output valves, which means a 5x5x4 tank, minimum. That’s what I built, and it adds substantially to the bulk of the reactor system.

    I had a really bad time with my thorium reactor.


    It just wouldn’t work right. The steam reserve kept fluctuating wildly, and the turbines would drift from 100% output to 10% or even stopped seemingly at random.


    I fixated for a long time on the steam delivery, but after far, far too much time going down blind alleys, I finally realized the problem was the large heat exchanger. The steam output was just all over the place, and wasn’t producing nearly as much steam as it should, overall.


    I’d slapped lots of one-way shutters into the steam pipes, which didn’t help. The moment I did the same to the hot coolant pipe, the turbines started delivering reliable power, and the steam buffer went from wild up and down swings to a slow, steady climb.


    I’m not sure exactly why the large heat exchanger is so sensitive to steady flow. You would think that even if the hot coolant delivery were irregular, it would average out to the same rate, but overall steam production was way down with irregular flow. It’s not like I hit the superheated steam threshold, either, since the small steel pipes can’t deliver 4000L in one second, even with sloshing.


    The final layout is a bit inefficient, because there are long (7-15 segment) pipes running steam from the buffer tank to the turbines. If I had put the turbines directly below the steam buffer, instead of on the same floor, I could have eliminated a lot of pipe. I did it this way because I wanted to be able to see everything easily, the reactor, exchanger, steam tank, and turbines, instead of having to run up and down stairs. It was really helpful when diagnosing my steam production problem.


    It doesn’t bother me too much because Huge PTFE pipes are relatively cheap and high throughput. It does make me wonder about the logistics of anything bigger and higher volume, though.

    Started up my thorium reactor. It’s apparent that my “steam expansion tank” was ill-conceived, it doesn’t behave at all how I imagined it would. Mainly because for it to work, the expansion tank has to empty really, really fast when the exchanger shuts off. The tank needs just as much throughput for dumping steam as a conventional buffer, so I might as well just build a steam buffer tank, the way I did for my boiler setup.


    I need to re-organize my reactor room. Which probably isn’t as big a task as it seems at first, though it does mean re-doing to the redstone logic controlling the various shutoffs.

    Here’s a pretty basic question:


    How would the non-technical people out there 1) learn that Linux exists, and 2) obtain a computer with Linux not just installed, but set up so they could easily use it? These are people who, as depressed_pho points out, have no idea what an operating system is.


    That “building it from source” is something you might need to do reminds me of CP/M. Back in the 70’s, CP/M was the “serious” operating system for micromputers, but it assumed that you were a programmer and comfortable with things like writing your own device drivers if you had to. When computers became a mass-market thing, sold to people who weren’t electronics hobbyists, it lost out to things like MS-DOS.


    You could install CP/M on an IBM PC or an Apple II, and in fact one of my earliest paying jobs was setting up a medical database on a Apple II running dBase II on CP/M, but it was never, ever going to become a mainstream OS.


    There is, of course, also the software ecosystem issue. Which is (probably) what Greg was referring to in his first post.

    And in the end people STILL use windows despite better alternatives

    What alternatives are you imagining? Mac OS? Because Linux doesn’t qualify, nor does any other Unix variant I’m aware of. It’s not remotely user-friendly enough for mass adoption. When you’re talking about “better alternatives” to Windows, you have to restrict yourself to OS’s my mother-in-law can use, or you’re not being realistic.


    I’m assuming you’re not including tablet OS’s like iOS or Android. They’re certainly usable enough, but they’re gimped compared to real desktop OS’s.

    Definitely. The reactor shell is no big deal, but all the support stuff with heat exchangers and turbines triples the amount of work involved in the block setup. Though there’s an argument to be made that placing all those internal components in the reactor is more work than the rest combined.


    I will say that nuclear control is really easy, and actually easier with a fluid reactor than a EU reactor, since it’s easier to place Project Red wires on the surface of a cube than on the cross-shape of an EU reactor. Simplest setup is monitor on an access hatch, NOT gate, AND gate with a lever, wire on a redstone reactor control spot. Or you can go with a NAND gate instead, but then the lever is reversed from what you’d expect.

    “My last couple of fluid reactors blew up” doesn’t exactly inspire confidence in your designs.


    Do people actually run reactors in survial without Nuclear Control safety shutoffs? I can see doing it in Creative. I’ve blown stuff up in creative because I cut corners while testing, but I can’t see doing it in actual play.


    I’m currently constructing my first survival fluid reactor, thorium based, and I intend 4 separate shutoffs for the reactor. Manual, temperature, excess steam, and distilled water shortage. I don’t think any but the excess steam will actually happen, but I can imagine screwing something up that might end up in a catastrophic runaway if I don’t have an automatic shutoff.

    I’m seeing serious water destruction in my boiler -> large turbine loop. I upgraded the rotors the 2x Large Ultimet and 1x Large Vanadium Steel, and the water just started steadily dropping. This isn’t a startup issue, it’s continuous operation. Steam production is closely matched to steam consumption (64,000 L/sec vs. 63,000), and yet the 2x Advanced distilleries could not keep up.


    It was so severe that I gave in and built a distillation tower just to make distilled water. It has no trouble at all keeping up with the water destruction, since it can output 626 L/sec, more than the combined consumption of the two boilers (426 L/sec distilled water). Still, I’m profoundly irritated.


    The turbines are receiving exactly as much steam as they need, I’ve got fluid regulators feeding them 27,000 L/sec for the ultimet rotors and 9,000 L/sec for the vanadium steel rotor. I’ve verified that all 3 turbines are, in fact, returning water while running. Just not enough to match boiler consumption.


    With my prior rotors, the system was stable at ~60,000 L of distilled water in the buffer tank during continuous load. By the time I finished the distillation tower, it was all gone. Fortunately I have shutoff logic on the boilers testing for less than 90% water in the input hatches.


    I’m 70% done with my first fluid reactor. All 3 turbines are built, as is the large heat exchanger and the reactor core. The containment vessel is 80% built. I’ve run no plumbing or control logic yet, and I don’t have a production facility for coolant yet.


    Rather than run all steam through a steam tank, I’m going to connect the steam output directly and attach an expansion tank to hold any excess steam, with some control logic to turn off the reactor if it gets full. It’s not really possible to exactly match steam production and consumption, though you can get close. Unlike the boilers, you can’t really throttle a heat exchanger. That just ends up backing up hot coolant, which is bad.


    I believe my production will be 76,000 L/sec and consumption will be 75,000 L/sec, but we’ll see how it goes in practice. I’ll probably tune the fluid regulators so the excess goes through the turbines, rather feeding the tank, but the tank will give me some leeway to avoid steam destruction (and thus water destruction).


    That is, if the steam turbines behave as expected. I don’t think they will, given what’s happening with my boiler / large turbine setup.

    I’ve actually got 3 HV blast furnaces, but only 2 of them are set up to transfer hot ingots to a vacuum freezer. #1 is my steel / general purpose furnace (hooked to oxygen), #2 is my titanium furnace (hooked to titanium tetrachloride), and #3 is my glow stone wafer furnace (hooked to nitrogen). I’ve got 3 because, in my experience, blast furnaces only take one fluid input. You can build multiple input hatches, but the furnace only checks one of them. Maybe that’s changed.


    Anyway, the idea of building additional furnaces makes sense. That 10 minute wait for tungsten’s ridiculous, and tungsten steel and HSS-G are what, 6.5 minutes? I remember looking at them but not checking the precise number.


    Once I’ve made HSS-G coils, I should be able to overclock tungsten and tungsten steel, bringing the time down to 2.5 minutes for tungsten, but HSS-G requires Naquadah coils to overclock. So building an additional furnace or two might still be a good idea.

    That’s a lot of reactors. Between the reactors themselves (5x5x5) and turbines, that must have taken up a lot of space. I’m guessing you only needed one heat exchanger even with that many reactors.


    My plan is currently to build just one fluid reactor with 7 quad cells putting out 950 heat / 2600 EU/tick from the turbines. I should make a point of automating the refueling, as you say.


    I may have overestimated how my power production is doing right now, because I noticed that I’m now short of wood. Usually I run a backlog from my multiple tree farms, but the conveyor belt and buffers are empty. The boilers are still running, but I don’t know for how much longer. Hopefully the diesel generators can keep pace if the boilers start stuttering.


    Power consumption is now around 2500 EU/tick, with peaks of 3000-4000. I’m not sure where all the energy is going, though obviously some is running the blast furnace continuously, and some is to running tungsten electrolysis which is 2000 EU/tick. Though that’s not too often since it requires 7 tungstate to run, and it takes about 3.5 minutes for my ore processing line to accumulate that much.

    Speaking of reactors...


    I think I finally have a theoretical understanding of reactor design, rather than just blindly copying internet designs. It helped to actually run a reactor and observe where the heat appeared to be going. A big problem with the simulator for me is that it tells you the result, but doesn’t single-step through the process so you can see why a design failed or succeeded.


    Overclocked heat vents have the highest heat dissipation at 20 heat. The best way to get heat into the OC’d vents is to just dump it into the reactor casing, since the OC’d vents draw 36 heat / reactor tick from the casing. Using heat exchangers to move heat to the OC’d vents is always going to be less efficient.


    OC’d heat vents will burn if you don’t eliminate the extra 16 heat. The simplest and best way is to surround each one with 4 component heat vents. Component heat vents are the second most efficient vents if fully surrounded, dissipating 16 heat each, better than Advanced heat vents which only dissipate 12. Thus the most efficient layout is always a checkboard of OC’d vents and component vents, averaging 18 heat dissipated per square.


    This breaks down at the edges of the grid, or adjacent to fuel rods, where it’s not possible to surround OC’d vents. There, a OC vent | advanced heat exchanger | advanced heat vent | component heat vent pattern is the best you’re going to do. The advanced heat exchanger draws 8 heat from the OC vent and 8 from the reactor shell, gives 16 to the advanced heat vent, where it’s dissipated by the advanced heat vent / component heat vent combo. The component heat vent dissipates an additional 8 from adjacent OC vents, so the total is 44 heat dissipated in 4 components, or 11 average.


    I don’t think anything more efficient is possible if you’re making a Mark I design that’s stable. I’m not touching unstable designs or MOX.