Alright, so: MOX reactor designs.

1st Official Post

    Looking for MOX reactor designs? Click here to go to the list!





    In theory, MOX fuel:
    - has a lifetime of 5,000 seconds, half as much as uranium
    - has exactly the same base EU output, heat output and efficiency scaling values as uranium
    - generates more EU the closer to 100% heat the reactor is, up to x5 at 100%
    - generates more EU the higher the reactor's actual heat level is (scaling factor unknown, but smaller than heat%) (was a bug, got fixed)


    The takeaway:
    - 100% reactor heat is, shall we say, "slightly impractical". A more realistic number is x4.4 at 85% heat, or x3.8 at 70% heat (no hurt)
    - Neutron reflectors last two times as many cycles and yield much higher total EU bonuses, potentially making them cost effective for once
    - The reactor must maintain its heat level exactly in order to give good results; this can be harder than you think


    Why can it be harder than you think? Because even designs that on the surface have cooling power equal to heat output will actually be unable to run at heat levels greater than 0. That is because you also need to look at a number the planner won't tell you anything about, and that is core transfer rate. Many reactor designs involving overclocked vents can actually pull more heat from the hull per tick than they are capable of dissipating. This works fine so long as common uranium reactors run at zero heat, but start the system off with several thousand degrees and the cooling system will melt itself even though it seemed stable in the planner.


    The following components have core transfer rates:
    - Heat exchanger, 4 core transfer
    - Advanced heat exchanger, 8 core transfer
    - Reactor heat exchanger, 72 core transfer
    - Reactor heat vent, 5 core transfer
    - Overclocked heat vent, 36 core transfer


    The reactor vent has 5 cooling, and thus will keep itself perfectly balanced on its own. The three exchangers will only utilize their transfer rates in order to make their own temperature (and that of adjacent components) match the reactor hull. The overclocked vent, however, will always pull the full 36 if it can, and it only has 20 cooling of its own. It will melt itself given the chance. However, so long as we know these transfer rates, we can use these parts build a reactor that satisfies the requirements for running MOX fuel.


    The simplest reactors, of course, get away without using core transfer at all: sample 1, sample 2 from another thread
    They simply have the fuel rods transfer heat directly into an adjacent component that can accept heat. If such a component is available, the fuel rods will never transfer heat to the core. Thus the reactor will maintain its heat level perfectly, and all we have to worry about is to provide enough cooling. The downside, of course, is that you're either running fairly low efficiency numbers (every fuel rod side occupied by a cooling component is one you can't use for another fuel rod or a reflector), or you'll invariably need a CRCS system to refresh spent coolant cells. Also, advanced vents are very expensive.


    The other option is to deal with core transfer rate: sample 1, sample 2 that I just made to illustrate the point
    This results in much higher efficiency reactors, with the downside being that efficiency means heat, and heat means you need to spend a lot of room on components. Which means less room for plating, which means less total heat capacity, which means less of an output multiplier. The trick question here is whether or not there is a point at which you lose more output by removing heat plating than you get from building a higher efficiency reactor. Absolute heat level seems to be a much smaller multiplier than total heat percentage, but it needs to be examined.


    Another downside of this approach is the fact that if you switch the reactor off (or the fuel rods run out), then the reactor will cool itself down to 0 unless you remove all cooling components or swap the fuel rods quickly. If you're slow, you might end up having to re-heat the components, or even the whole reactor, for the next cycle.


    Design Q&A:
    - why does sample 1 work, the core transfer is higher than both heat output and available cooling! Answer: because the reactor heat exchanger only draws as much as it needs to keep itself at the same temperature as the hull. That means it will draw exactly as much as the cooling components steal from it each tick, which is exactly as much as the fuel rod pumps into the hull each tick.
    - why does sample 2 work, it has more cooling than heat output! Answer: because it only has enough core transfer to pull 168 heat per tick from the hull, even though the cooling system could handle 172. If it bothers you, you can move one of the component vents so it only touches one component instead of two... but then you lose THE SYMMETRY! MinecraftCow



    Please share your thoughts on the subject, and most importantly, your design ideas. Let's come up with a few that are worthy of a new section in the Official Reactor Design thread!

    Another often underused component which has ZERO core heat exchange is the Component Heat Exchanger. All it does is pass heat back and forth between components, meaning that you can pass heat off to other components without risking altering the temperature flow of the actual MOX design.


    Furthermore, each component heat exchanger can transfer 36 heat per tick. Now, here's some math for you:


    Max heat transfer for a Component Heat Exchanger: 36
    Max heat dissipation from an Advanced Heat Vent: 12
    Number of sides remaining on a Component Heat Exchanger next to nuclear material: 3
    3*12 = 36


    Meaning that one component heat exchanger can keep things running smoothly with three advanced heat vents adjacent. Basically giving you 36 heat dissipation without otherwise harming the delicate temperature equilibrium of your reactor.


    As a not-very-economical demonstration, I present the Heat Distribution Transfer Vector reactor. The HDTV takes advantage of this curious mechanic to handle a pair of quad MOX cells in a 3 chamber reactor. The 'excess' cooling is in the component heat vents, so it will not impact the overall core temperature.


    Another interesting design concept is to incorporate Cooling Cells which are then cooled off through various means as a way of transferring heat from nuclear material on out without upsetting the balance of the heat in the reactor itself, although not in a CRCS setup. here is an example of this mechanic in practice.


    Of course, I still maintain that the CRCS reactors are going to be the stars of this show. The Pocket Split DDoS reactor would only see an increase in efficiency and EU output from using MOX fuels at high temperatures, with zero additional drawbacks, and with NR's being more economically viable, the old Alpha Pocket DDoS reactor might get dusted off and re-commissioned into service. Or, for those of you who can handle the output and simply don't want to be satisfied with that, the original Alpha DDoS reactor in all its glory.


    ALL CRCS reactors would get a significant boost from MOX. In effect, it's giving them free efficiency without otherwise changing any of their previous engineering challenges. So if you have a stable CRCS reactor going, there really is no reason to not MOX it up once you get the materials at hand.

    The issue with CRCS though is that they become difficult to automate in the absence of certain mods. As such I keep being hesitant about the idea of recommending them to the broad public.


    Fun idea with the concept of supporting coolant cells with vents. Toyed around a bit, came up with this: http://www.talonfiremage.pwp.b…on9lt7xgoqiv37kb0brx9nyy0
    It's got a decent heat capacity, it has a good efficiency, it will work with single cells and quad cells too, and running it with dual cells won't spontaneously combust your bank of MFSUs (like quad cells certainly would). The 11 minute burst allows for a lot of power in a short time, after which the reactor can cool down. It almost, but not quite, manages a full cycle with single cells.


    On the downside, the cooldown time is so massive that after running for 11 minutes in the dual cell config, you have to let it sit for 48 minutes to cool. Those few puny vents, expensive as they are, just don't really cut it, and the result is an effective EU/t of only around 17% of its actual rating. Provided you don't swap the coolant cells out, of course. But if you do, then you might as well go CRCS. Overall, I'm not sure that design is the way.


    Some other food for thought: not only reflectors, but also condensators become more cost effective: http://www.talonfiremage.pwp.b…styj6onfrmpbxg0ifwnzzo2g4
    Conventionally we'd be looking at 240 lapis for 400 EU/t here, a rating that even non-condensator reactors can reach with internal cooling (albeit at the cost of uranium efficiency). But after the MOX multipliers, this thing should be putting out in the neighborhood of 2000 EU/t (a minimum of 1750 for running it at 85% temp, plus whatever bonus 35k absolute temperature adds). As you may remember, Direwolf20 built this to achieve a similar rating - at the staggering cost of 1371 lapis per cycle, instead of 240! (Note: MOX cycles are half as long, so it's 120 lapis per cycle, but over identical time the consumption speed is identical to uranium).


    Still not sure if I'd want to run that - the issue of condensator replacement remains, since condensators can handle even less than coolant cells before needing replacement actually they can handle more, but it's a difference of 8.5 to 11 minutes in the above reactor, which is still not much at all. But at least this means we now no longer need to yell at every newbie who comes here with dreams of gigantic condensator fests... MOX fuel actually gives them a reason to exist.

    Okay, just tested the condensator reactor ingame. Good grief - with zero plating and at 8,400 / 10,000 heat, this thing produced 1,932 EU/t. With full plating and at 35,632 / 42,000 heat, it was pushing 2,915 EU/t. Finally, I bumped it up to 6 chambers and tested again at 66,352 / 78,000 heat: 3,331 EU/t. Definitely seeing diminishing returns there, but damn... that's more than 8 times the EU/t of the uranium variant, still with identical fuel layout and identical lapis consumption.


    CRCS should see similar gains, considering the only difference is using coolant cells instead of condensators.


    Still looking for a good and efficient internal vent reactor though...

    Well, advancent vents are the weapon of choice in MOX reactors, after all - if you want maximum cooling without core transfer. And, I wouldn't worry so much about that one diamond per vent, the rest of the vent is also noticably expensive now that basic vents require motors :P Of course, at the time you can afford MOX fuel rods, you should be well covered on the resource front. Compress some coal if necessary.


    Now, I've been testing ingame, and much to my disappointment, MOX fuel seems quite buggy: EU output is highest when the fuel cells are closest to the left side of the reactor, and declines significantly the further right you move them, even if the design remains the same. Observe:


    In my search for a decent efficiency, internal cooling MOX reactor, I wanted to test this design. In between taking various measurements with various amounts of plating and reactor chambers, I noticed the aforementioned bug. Note: I did make a silly mistake ingame, in the form of forgetting the two component vents in the design, but that didn't influence the test. Reactor temperature for the following five scenarios was always 59,520 / 59,500 / 70,000, minimally overshooting the melting point but irrelevant for testing purposes.


    Scenario A: 578 EU/t
    Scenario B: 552 EU/t
    Scenario C: 529 EU/t
    Scenario D: 508 EU/t
    Scenario E: 489 EU/t


    My first reaction: "What is this I don't even..." :wacko:


    This more or less means that while we can try to come up with designs under the assumption that things will eventually scale correctly, getting accurate numbers on how any given design actually performs ingame seems impossible right now. ?(



    EDIT: submitted bug report on this.

    I registered to chime in on this issue. I've had a peek at the code so I can clarify things a bit, hopefully without spoiling all the fun.


    Firstly, the snippet of code Thunderdark posted in the other thread doesn't seem to be omitting anything. I think the weirdness you're seeing is a result of execution flow. As you may already be aware, each component in the reactor is evaluated sequentially, starting from the top left slot, moving left to right, top to bottom. For example, this behavior is the reason that in some reactor designs overclocked heat vents at the top are more susceptible to overheating than ones at the bottom. This execution flow also influences MOX behavior, as the "heat effect" is calculated per MOX cell before the MOX cell adds its EU contribution to the reactor total. It's not unlikely that the heat level of the reactor will change between MOX cell evaluations, even within the same reactor tick. So even if the reactor is balanced and it's heat stays "constant" from tick to tick, the heat does change during the tick.

    Unfortnately this does not explain the issue I am seeing.


    For starters, what you wrote is true for reactors that exchange heat with the hull. Pretty much every normal uranium design does this. First, the cells dump heat straight to the hull, and second, the cooling system pulls it back out. That's possible with MOX reactors too, as I outlined under option 2 in the opening post.


    However, the reactor I was testing with in the post directly above yours is a type-1 MOX reactor. It has zero core transfer rate. The cells dump their heat directly into heat exchangers, and the vents take it directly from the heat exchangers. The hull is never involved, and hull heat remains rock stable.


    Secondly, even if it was a type-2 MOX reactor, the only thing that changed in the layout between the five test runs were the heat platings, which are never evaluated for any heat mechanics beyond increasing the maximum hull capacity. The rest of the setup is preserved exactly, just shifted to another position. This means that the evalutation sequence is also preserved exactly. There should not be any difference in how much heat is in the hull at any specific point in time between any of the setups above; hence, they should provide identical output.

    While reading about MOX reactors is interesting, how exactly do you get the reactors up to these high temperatures without heating cells? Equally important is how you stop them overheating once you've got them nice and toasty.

    Quote from Shakie666

    While reading about MOX reactors is interesting, how exactly do you get the reactors up to these high temperatures without heating cells? Equally important is how you stop them overheating once you've got them nice and toasty.


    The way to do it is fairly easy, if you are using a DDoS MOX reactor you keep the cooling cells away from the MOX cells, then when the reactor gets to a good temperature, you move the coolant cells next to the MOX cells.


    If you are using core exchange and vents, then you want to have the vent cooling be exactly that of the MOX heat output, keep the vents out of the reactor then when the reactor is hot, turn it off and add in the vents.

    Type-1 reactors are fairly easy to heat. Unfortunately, having the Nuclear Control addon installed is pretty much mandatory at this point, since you need a way to measure temperature.


    I build an industrial information panel, hook a sensor kit up to the reactor, and then place in all the plating that I want to use. Then I just use some kind of reactor fuel to inject heat - quad uranium cells work pretty well, especially if you group them up, but you probably want to keep something smaller around for the last few degrees. I bring it as close as I can to the melting point as indicated by the info panel. Then I insert the actual setup and MOX fuel. Since it doesn't affect the hull heat at all, you can stop worrying about it.


    Type-2 reactors are much more difficult, though. You need to bring the heat up, but not quite as close as before. Then insert the MOX fuel cells and the components without core transfer. Then, after you switch the reactor on, rapidly insert all the core transfer relevant components. Start with the ones there are the least of; ideally you can shiftclick the type with the largest amount in last, courtesy of inventory tweaks.

    *mod-setup* w/ nuclear control.


    so i just tossed two mox cells in a reactor with a reactor vent. the cells are vertical-adjacent in the bottom with the vent at the top of the reactor. scram at 8k (redstone torch 'nor' gate). reactor is cranking at 84 eu/t 2 on, 4 off second intervals.
    added 3 more R-vents... 6 on, 2 off, same output.
    since i was taking damage, i lowered the scram to 6990(threshold). output dropped to 76 eu/t...but is serviceable.
    the concept of this setup was that i could let the reactor heat up with no cooling, it would turn off at temp, i put in the 4 vents, and it will run fairly efficiently. it would also be a cheap(er) build.
    i do not recommend using the remote detector, as it is very energy hungry while the thermal monitor doesn't require any energy, and since reactors are no longer cooled by surrounding water (i checked, not re-implemented), having it adjacent shouldn't be an issue.
    this obviously would not go over well in the 'official' reactor design thread, but these things run on automation IRL for a reason...

    Brainstorming about how to express efficiency of MOX reactors.


    This gets a bit silly, since you have what effectively comes out to three two different internal efficiency values multiplying another for the final number. You have the efficiency that comes from fuel cells having neighbours to bounce off of in the setup, just like uranium. But you also have the efficiency that comes from how high percentage wise the reactor heat is; in other words, how close you are to blowing up. The closer to doom, the better. And then in addition to that, you also have the efficiency that comes from how hot your reactor is, absolutely. The hotter the better. (that was a bug)


    Currently I'm leaning towards calculating it via EU/t, since that's a metric the player has easy access to, either via the eu-reader (when it works again) or via the industrial information panel (because if you want to use MOX reactors you need to install Nuclear Control anyway).


    Step 1: Add up your number of individual fuel cells; a dual cell obviously counts as 2, a quad cell as 4
    Step 2: Multiply by reactor base output as defined in your config file. Default 5, is equal to the output of a single fuel cell at eff 1
    Step 3: Divide your recorded EU/t output by that number


    Example: I have a reactor outputting 577 EU/t from two dual MOX fuel cells, two reflectors and tons of plating. My IC2 config is set to reactor base output 6.
    ---> 4 cells * 6 EU/t = 24 EU/t
    ---> 577 EU/t / 24 EU/t = 24.04


    If my reactor base output was 5, as the default, I would be seeing an output of 481 EU/t, resulting in the following math:
    ---> 4 cells * 5 EU/t = 20 EU/t
    ---> 481 EU/t / 20 EU/t = 24.05


    And the results are comparable to uranium reactors, since the formula applies to them too. Take this popular design:
    ---> 6 cells * 5 EU/t = 30 EU/t
    ---> 100 EU/t / 30 EU/t = 3.33 ...which is exactly what the reactor planner says.




    To sum up, this method gives us a number that is directly comparable to uranium reactors and is (within a tiny margin of error) independent of the user's reactor base output config, eliminating mistakes like I just love making them all the time. ;)


    What do you guys think? Can you come up with any flaws in my train of thought? Are there any edge cases where this doesn't work out that I might have missed? Got a better idea to measure MOX efficiency? Let me know!




    And I'm still keeping my fingers crossed for a bugfix so we can start making designs for real... :sleeping:

    We have this one in use:
    21p7h4fnat72neq5m0brbcfwkakqknk1reeycb7ppun1hj9iq897n5i9f7wyog6wju93sib6kfv92ce
    Thick Neutron Plating is too expensive atm as the lack of iridium.


    So (default config | T=79888 :(
    -> 783EU/t / (8 cells * 5EU/t)= 19,575


    Looks good :)

    Well, if both absolute -and- relative heat levels are a factor in MOX output, then you're going to see a bunch of six-chamber reactor designs filled with heat plates in spare slots.


    The interesting thing here will be the cost effectiveness of that design. Will it be worth the price of extra chambers and the copper in the heat plating to increase the size of a reactor just to have a higher cap and current heat value? Where will the diminishing returns be?


    MOX reactors will need to be approached differently than standard reactors. Things which depend on drawing heat from the reactor itself (such as having a quad of cells in the top corner then a bunch of alternating OC vents and item heat vents) won't work. You're going to be fluctuating too much and might end up with unpleasant effects accidentally. You're going to want something which depends primarily on side-to-side heat exchange, which puts some rather sharp limits on how efficient the reactor can get, since you are sharply limited by how much heat you can draw out of a single square in the reactor.


    Of course, CRCS designs just totally blow that out of the water. There's absolutely zero reason to not MOX up a CRCS reactor (other than lack of materials to make the MOX fuel). But then... CRCS has its own problems and issues which will need to be solved.


    I'm almost wondering if there is a way to make a hybrid CRCS system which uses cooling cells to pull heat out of the fuel by contact, partially cooled in conventional means, and swapped out maybe once per cycle. That would be a far less dangerous way of running a MOX reactor that might produce some very interesting results.


    Take, for example this reactor design. It has a micro-cycle time high enough that you can get away with a single set of cooling towers if you use towers with a cooling of 120/cell/tic or greater.


    Normally not a very efficient CRCS design, since the split-tower does a better job, it does have a longer micro-cycle time which makes automation much easier.

    I use overclocked heat vents that are transferred to a cooling tower once they hit half-life. Each heat vent cools the reactor by 36, and cools itself by 20, which leaves 16 remaining. Because 16 is less than 20, they cool themselves in the tower quicker than they heat up in the reactor. I have a filter to withdraw them from the reactor only at half-life as well as one (16 unit) step in either direction and also at the last 6 steps before it melts. another filter removes it from the cooling tower at full health and stores it in a chest. Finally a filter moves them from the chest into the reactor. Using a combination of overclocked vents and regular reactor vents, I can usually get it pretty close to 0 heat gain/loss.


    I also use nuclear control to keep it in its heat range, but that is kind of cheating. I'm sure with a bit of effort a redstone timer could be built to keep it within its bounds.


    I've never seen a similar method used before, so I'm curious what you guys make of it.

    Quote from JaxFirehart

    I use overclocked heat vents that are transferred to a cooling tower once they hit half-life. Each heat vent cools the reactor by 36, and cools itself by 20, which leaves 16 remaining. Because 16 is less than 20, they cool themselves in the tower quicker than they heat up in the reactor. I have a filter to withdraw them from the reactor only at half-life as well as one (16 unit) step in either direction and also at the last 6 steps before it melts. another filter removes it from the cooling tower at full health and stores it in a chest. Finally a filter moves them from the chest into the reactor. Using a combination of overclocked vents and regular reactor vents, I can usually get it pretty close to 0 heat gain/loss.


    I also use nuclear control to keep it in its heat range, but that is kind of cheating. I'm sure with a bit of effort a redstone timer could be built to keep it within its bounds.


    I've never seen a similar method used before, so I'm curious what you guys make of it.


    This sounds like an HVC system, which has a very serious problem with micro-cycle timing needing to be -exact- or you get a reactor explosion. Also, your calculations sound very off. Nuclear components can generate over a hundred heat per tic which is absorbed by adjacent components.


    If you are talking about pulling from the hull itself, that carries another very serious danger when you are operating at an already high heat with almost zero safety margins. One slip-up, and you have a crater guaranteed to give any creeper explosion-envy.

    Yeah, I'm talking about pulling heat from the hull. Yes it does allow for meltdown potential, but I use nuclear control to keep the heat within an acceptable range. Because I'm pulling heat from the hull, the numbers are real easy: So long as the Overclocked Heat Vents pull the maximum 36 heat from the hull, they cool 20 of it and absorb 16 heat. Because I know heat is accrued in steps of 16, using a filterable item-moving-mod (I use CodeChicken's Translocators) I can tell it to automatically withdraw them at half heat capacity. So instead of using redstone timers, I use specially configured filters. These are not vanilla ic2 Mox reactor designs, but are fairly minimalist. I have no need for Component Heat Vents or Exchangers. I simply remove the Heat Vents at half capacity and let them set in an empty cooling tower to cool of. So each Overclocked Heat Vent provides a massive 36 cooling and every slot in a reactor can be occupied by either heat plating, reflectors, cells, or vents.


    http://www.talonfiremage.pwp.b…7h6y7kyqphjsl6kfs60z6sirk


    That reactor has a perfect 7 in efficiency, and generates 448 heat. 5 of that heat is dissipated using a standard reactor heat vent. 432 hull heat is absorbed by the 12 remaining Overclocked vents, of that heat, 240 is dissipated, leaving 192 remaining, each overclocked heat vent gains exactly 16 heat. This is essential. Because filters are metadata specific, you HAVE to ensure that the components accrue the same heat at the same rate constantly. That's why I went with 12 Overclocked and 1 Reactor instead of 13 Overclocked. 12 + 1 results in a build up of 11 heat, whereas 13 results in in a net decrease of heat. 13 also results in one of the Overclocked Heat Vents having a build up of less than 16 heat each tick, which makes the filters useless and leading to a melting heat vent. The second they hit half capacity they are transferred to a cooling tower and fresh ones are brought in. Using translocators, at least, this is instant. Finally, nuclear control ensures that, should something go wrong, the reactor won't go critical and is also used to shut the reactor down once that 11 heat gain manages to build up to dangerous levels.

    Quote from ShneekeyTheLost

    Well, if both absolute -and- relative heat levels are a factor in MOX output, then you're going to see a bunch of six-chamber reactor designs filled with heat plates in spare slots.


    The interesting thing here will be the cost effectiveness of that design. Will it be worth the price of extra chambers and the copper in the heat plating to increase the size of a reactor just to have a higher cap and current heat value? Where will the diminishing returns be?


    MOX reactors will need to be approached differently than standard reactors.


    This is something I'm raring to find out about. For example, I made these three reactors for the express purpose of comparing the value of extra plating to the value of extra cooling for higher efficiency and/or output.


    But I can't be confident in my findings until the location-specific output bug is fixed. I'm also pretty sure that we'll see different output numbers then for our existing MOX designs, so trying to judge cost effectiveness now is pretty much an exercise in futility.



    Regarding the efficiency calculation proposal, I remembered last night that MOX fuel only runs half as long as uranium fuel. Now I'm not sure if this should be significant... I mean, the EU/t difference is what it is, no question, and so long as you have fuel available you have an effectively permanently on reactor in either case, ignoring cycle times. But technically, if you were using efficiency in order to try to describe what each single fuel rod can do for you, and you wanted full comparability to uranium reactors, then you'd have to go by total energy created per cycle, not EU/t. You can't directly see that number anywhere ingame, but conveniently the runtime difference of one half means that all you need to do is divide the MOX efficiency numbers by two after the calculation presented further up.


    That would result in an efficiency of 12, not 24, for the example numbers I used. Which is, admittedly, still a darn good number - three times as high as many of the best balanced uranium designs.


    The question is, should an effeciency 4 uranium design that gets its fue rods replaced with MOX but remains otherwise unmodified automatically get downgraded to eff 2, because the running time halved? Or should it still remain efficiency 4, because that's what the cell configuration is?

    In my opinion, efficiency is a measure of how well you are using your fuel, so while cycle time and start up costs (getting to MOX etc) should be considered by potential builders, the efficiency level should not change based on what fuel rods are used.