[WIP] Nert's Dev Thread - Current: such nuke, wow

[DerekL1963]

Did some testing of NFEX_4_10 with various configurations of LV-N's and the monster LV-NB (from KSPX) last night. Seems to be a reasonable balance, the radiators are heavy enough to give a noticeable performance hit (accounted for easily enough by either attention to your design or MOAR BOOSTERS!!), but perform well enough that you don't need to spam them in mass quanta to keep temperatures within a decent range for any normal burn length.

Followed up tonight by full up tests with NFEX_4_11 and an actual craft (stock other than NFEX radiators) braking into Moho orbit on the dayside (one of the two longest burns and worst thermal enviroments the average player is likely to encounter). Doable, but it does take attention to thermal design and the vehicle's available D/v (as piling on too many radiators causes too great a performance hit). Working as intended I suspect, and it feels balanced.

For reference, the vehicle used for the test; Four radiators worked, but I went back with eight because the four radiator version seemed marginal thermally.

Tomorrow, sundiving into low Kerbolar orbit for sweet, sweet science. (This version failed sundiving during the long burn (out near Kerbin's orbit) to lower the Pe.) Too tired to work that problem tonight, after I hit "submit" and have a smoke, it's bedtime for me.

[Streetwind]

Okay, I've gotten to the bottom of the issue with the weird heat conductivity I've been seeing. Turns out that the heat insulator parts are not involved after all.

What happens is that KSP fails to read the heatConductivity stat from the part config when it puts the rocket onto the launchpad. That means that the radiators get loaded with the game's default conductivity values.

During flight, whenever you stage, there's a small chance that it'll update itself, and load the proper heatConductivity value for the radiators... which is very small. Thus, their conductivity drops dramatically. The chance for this to happen seems largest when the stage the reactor is in is activated, but it's not guaranteed to happen.

You can force the correct values to apply by quicksaving and quickloading. This results in the correct values 100% of the time.

Kind of a lousy bug, because it really affets the performance of NF reactors and radiators to a large degree But at least I can now replicate it reliably, which means it's time to toss a bug report Squad's way. They're working on a hotfix patch this week, or so Maxmaps claimed, meaning this is a great opportunity to get this fixed. Nertea, if you still hang out with Roverdude in chat sometimes, maybe you can poke him about it?

EDIT: currently trying to reproduce this with a stock part for the bug report... and getting stumped. It does seem to be loading things correctly for stock parts...

Edited May 14, 2015 by Streetwind

[MatterBeam]

The real problem is the wing's thermal tolerance is too high... 2000K radiators are a carbon metamaterial magic (that is possible and does exist) and would be used on a reactor like the SAFE400. You'd heat those radiators up so far that they'd emit IR energy like crazy and be very efficient even at a small size. And yes, you'd use heat pipes to get to them, then the coolant would go back into the radiator to heat up the coolant again, and repeat. This is how electricity is generated, it's an ideal engine cycle (Brayton Turbine in the case of the SAFE400).

Interesting thing is when you heat a radiator up that far they don't need to be that big either, the hotter the radiator's temperature, the more kWt it loses to space. (the increase is to the 4th power)

The reason the ISS needs such big radiators is they aren't very hot.

In this case, even on the ground, most of your heat loss is through radiation not convection or conduction. Actually being on the ground means that you're radiating at a grey body, so it's not as efficient. In this case facing out into black space, yep... this is working as it should from what I can tell. Those two radiators facing the sun shouldn't be radiating as much, but those two perpendicular to it would be very effective.

On the radiators for the NTRs: I haven't been watching this thread super close, but the NTRs should be losing all of their bulidup heat through the propellant while running. It's when they're not running that something must be done to cool the reactor (which could be done by turning down the fission reaction, which means you would have to spin it up again before you could use the engine again). Probably using it as a generator to turn those watts of thermal energy to electrical energy, then you'd radiate your heat energy off into space. Basically using it like a giant RTG, large radiators should not be needed. I've yet to fiddle with the stock NTR... but from prior use and other engines it sounds like it's backwards. It generates a lot of heat while running and is cool when it's idle. The heat is what makes it work and have its high ISP!

AFAIK, there's several things wrong here:

-Conduction through dense atmosphere, such as on the launchpad, is many many times more efficient than radiating, at all temperatures.

-2000K is about 1726 degrees Celsius. That's well within the temperature tolerances for many solid materials. While a command pod shouldn't be sitting at temperatures that melt steel, there's no reason a carbon radiator doesn't operate at temperatures well below its triple point in atmosphere (4330C) or its sublimation point in vacuum (5530C).

-NTRs are solid core rockets. Their temperature can increase or decrease depending on how far the neutron poison control rods are inserted/rotated. There is no reason that propellant flow is always enough to cool the nuclear core. While it might be true for chemical rockets, because of their massive fuel flows (up to 1.2 tons per second), it is not true for the LV-N which only has a fuel flow of 7.6kg/s to dump its heat into.

-Converting waste heat into electrical energy does not make that waste heat disappear. You'll always need to dissipate the full amount.

[Streetwind]

Bingo! Issue identified!

Near Future radiators have low heatConductivity values only when deployed. When folded up, they use the default conductivity. However, they do not update their heatConductivity value when toggling.

That is why the radiators get loaded on the pad with a high conductivity, which they then keep even when the radiators are deployed. And updating them with staging or quickloading only works if they are in a different state than before; there's nothing to update if they are stil lin the same state.

Turns out this was a NF bug after all... =/

Edited May 14, 2015 by Streetwind

[DerekL1963]

Tomorrow, sundiving into low Kerbolar orbit for sweet, sweet science. (This version failed sundiving during the long burn (out near Kerbin's orbit) to lower the Pe.) Too tired to work that problem tonight, after I hit "submit" and have a smoke, it's bedtime for me.

Over coffee this morning, I realized why my tests with _10 seemed so easy, and with _11 so hard. In _10, I tested what I assumed would be ruling case, my heavy Duna tug with either multiple LV-N's or one monster LV-NB... but that vehicle is based around a 2.5 meter hull and has a much larger thermal mass. The Moho probe and the Sundiver are based around 1.25 meter hulls.

Anyhow, a successful 20 minute burn (out near Kerbin) and a successful Sundiver (survived and was able to gather low Kerbolar orbit Science at a peri of 900Mm);

In reality, looking at the temperature overlay and gauges, in a "real" game (as opposed to sandbox testing) I'd be sore tempted to toss on a couple of additional XR-500's as near the end of the burn and at peri there were a number of components operating just shy of redlining. Overall, it still seems balanced, it's easy enough to build a vehicle with sufficient performance and thermal controls if you pay attention.

Edited May 14, 2015 by DerekL1963

[Streetwind]

Examination of radiator performance.

The standard radiators, XR-500 and XR-2000, are quite well dialed in. If you ask them to dissipate the entire 550 / 2250 kW that they can draw, they stabilize at 1225 K and 1180 K, respectively. At that temperature they have a nice bright red glow, with the XR-500 starting to lean towards orange-ish.

Though perhaps their names could use adjusting, considering that the capacity numbers changed...

The conformal radiators, as it turns out, have more surface area than they need for the amount of heat that they are configured to handle.

The GR-4 is the hottest candidate, stabilizing at around 1060 K. At that point it possesses a decent, if unexciting, middle-red glow.

The GR-1 doesn't get that hot, even if dissipating 1350 kW instead of the strictly required 1250 kW. It'll stabilize around 940 K, at which point it has a dull reddish glow.

The GR-EXP, finally, is completely bored by its task and stabilizes at no more than 825 K. Its glow is so weak, it's barely visible even in the dark.

I also tested with emissiveConstant of 0.95 instead of 1.00, which generally increased the stabilization temperature by 20 K to 30 K. This didn't negatively impact any of the radiators. Even the XR-500 running at 1250 K is still in no danger of melting.

With the deployed radiators actually getting their lower conductivity after quicksaving/quickloading, the radiators also have a very generous breathing room in terms of heat transfer capacity. For example, with the "wrong" conductivity the GR-4 has so much backflow that the whole ship stabilizes somewhere a bit above 400 K, and the radiator at 225 kW transfer capacity is well balanced to keep it there. But with the correct conductivity, the radiators have no trouble cooling the entire ship to 273 K (below which they do not cool) and keep it there with performance to spare. Even adjusted to 210 kW capacity, the radiators are still easily sufficient.

Now obviously, engines also produce heat. The overcapacity of the radiators helps cool the ship during and after burns. Depends on your tastes how much overcapacity you'd like to have.

Finally, I tested just how much room for capacity the conformal radiators actually have with their areas as they are. With emissiveConstant at 0.95, I tried to get them to 1200 K, which is roughly the same heat range as the standard radiators.

The 55 kW GR-EXP can dissipate 220 kW at a temperature of 1200 K, with negligible conductive backflow

The 225 kW GR-4 can dissipate 300 kW at a temperature of 1195 K, with minor conductive backflow of about 3.5 kW

The 1350 kW GR-1 can handle 3300 kW at a temperature of 1199 K, of which 125 kW is conductive backflow and 3175 kW is radiated

I was pretty surprised by how well the GR-EXP performs, and how little difference there is to the GR-4. I mean sure, the former is meant to be mounted in sets of 4 and the latter in sets of 6, but still. Interestingly enough, the heat capacity estimation in the VAB (which I still have in my version) mirrored this from the start, showing the GR-EXP surprisingly close to the GR-4.

Edited May 14, 2015 by Streetwind

[smjjames]

Is this stable enough? I want to use it to keep the nuke engines that I have from overheating. Sounds like there is a bug regarding them being toggled or something.

[billkerbinsky]

-NTRs are solid core rockets. Their temperature can increase or decrease depending on how far the neutron poison control rods are inserted/rotated. There is no reason that propellant flow is always enough to cool the nuclear core. While it might be true for chemical rockets, because of their massive fuel flows (up to 1.2 tons per second), it is not true for the LV-N which only has a fuel flow of 7.6kg/s to dump its heat into..

Sure, but in a real reactor (not yet modeled in KSP), there will be decay heat (from the decay of unstable fission products with half lives measured in hours to years) even after the main chain reaction is stopped via the control rods. That initially starts at around 6% or so of the reactor power for typical U235-burning reactors but decays exponentially.

[Fraz86]

@Streetwind, regarding your examination of radiator performance:

Your findings are interesting, however, I'm curious, do you think there's necessarily a problem? I'm not sure there is. The real value of radiators is in their active cooling capability. Of course radiators need to have enough dissipation to keep up with the flux from active cooling, but beyond that basic requirement (which they all satisfy), dissipation capacity is not really what you're paying for. As my do-it-yourself radiator experiment demonstrated, comparable dissipation can be achieved with stock parts for a fraction of the mass and cost of radiators. But that won't keep you under nominal temperature without active cooling. Thus, as long as the active cooling rates are well balanced across radiators and they're all able to dissipate the transferred heat, I'm not sure there is any cause for concern.

[Streetwind]

Is this stable enough? I want to use it to keep the nuke engines that I have from overheating. Sounds like there is a bug regarding them being toggled or something.

It's a dev build - there will be bugs. You can't really expect me to say anything but the responsible thing: wait for the full release. There's a chance it might happen this weekend.

If you're okay with quicksaving/quickloading everytime you extend the radiators, then I suppose you can try it though...

@Streetwind, regarding your examination of radiator performance:

Your findings are interesting, however, I'm curious, do you think there's necessarily a problem?

I'm not saying there's a problem, no. Nertea specifically asked for feedback on the conformal radiators and their intended function with their respective reactors. They are meant to come in specific sets that are mounted together, and those sets need to be functional.

I'm commenting on that, as well as anything else I noticed surrounding these parts. I'm hoping it will help Nertea make informed decisions about part balance. There is potential, for example, to stat up the reactors differently because the radiators can support a higher heat load. Whether this is necessary or desirable is up for discussion, but the potential exists. "Your findings are interesting" is exactly the kind of response I hope to create

If there's a problem anywhere, I'll point it out specifically and explain why it's a problem.

I mean hey, you can probably read between the lines in my post that I prefer brightly glowing radiators over dull glowing ones, but I'm not about to burn the house down over the issue

Edited May 14, 2015 by Streetwind

[Starbuckminsterfullerton]

I think the problem isn't a performance one, it's just a visual thing. It's nice to have radiators that look like they're doing something, and it helps tell you if you have the right amount. It's fine if they run cool, but then we'd never get to see the Nertea's beautiful emmissives.

That and it doesn't make much sense for the half meter conformal to do as much work as the on meter one.

Edited May 14, 2015 by Starbuckminsterfullerton

[PickledTripod]

there's a chance it might happen this weekend.

Hype Mode Activated!

EDIT: Propulsion+Electrical or just Electrical?

[helaeon]

AFAIK, there's several things wrong here:

-Conduction through dense atmosphere, such as on the launchpad, is many many times more efficient than radiating, at all temperatures.

-2000K is about 1726 degrees Celsius. That's well within the temperature tolerances for many solid materials. While a command pod shouldn't be sitting at temperatures that melt steel, there's no reason a carbon radiator doesn't operate at temperatures well below its triple point in atmosphere (4330C) or its sublimation point in vacuum (5530C).

-NTRs are solid core rockets. Their temperature can increase or decrease depending on how far the neutron poison control rods are inserted/rotated. There is no reason that propellant flow is always enough to cool the nuclear core. While it might be true for chemical rockets, because of their massive fuel flows (up to 1.2 tons per second), it is not true for the LV-N which only has a fuel flow of 7.6kg/s to dump its heat into.

-Converting waste heat into electrical energy does not make that waste heat disappear. You'll always need to dissipate the full amount.

From when I was looking into this earlier (and what I remember from my physics and chemistry courses in college):

In an atmosphere, that is not moving, you can saturate the air and have limited further conduction. Now this in the real world is usually not a big deal because the hot air will rise and carry heat away. In which case yes, the active cooling of the radiator by the external fluid (the atmosphere) will be incredibly effective. Far more than radiation against a gray body which you always are on when you're on the surface. I don't think KSP is modeling atmospheric convection.

Yep about 1700C, and you need to run the coolant through that, it also needs to maintain structural integrity. I was mostly looking into the SAFE400 and the temps it was operating at, 2000K required a carbon radiator. For the reasons you stated.

Are we controlling the NTR's core as such that we can over-do it and the propellant is not enough? Point of the NTR is to get it as hot as possible in a stable way and run it. For game play purposes, the propellant should be enough to get it done. And the control rods should handle it for when its deactivated. Much like prior KSP NTR behavior, that is unless we can crank that heat up to get an ISP boost understanding it will run for a limited time and we are responsible for that extra cooling.

Not waste heat, total heat. Some of the heat energy becomes electrical energy in the converter, rest is waste heat. Yes you do need to get rid of all of that. So if you have a 2000 kWt reactor, 400 kWe converter, you need to dissipate 1600 kWt to the environment somehow or otherwise use it... though limits there as you approach the ideal engine.

[Nertea]

Bingo! Issue identified!

Near Future radiators have low heatConductivity values only when deployed. When folded up, they use the default conductivity. However, they do not update their heatConductivity value when toggling.

That is why the radiators get loaded on the pad with a high conductivity, which they then keep even when the radiators are deployed. And updating them with staging or quickloading only works if they are in a different state than before; there's nothing to update if they are stil lin the same state.

Turns out this was a NF bug after all... =/

So I don't actually touch the conductivity ever in code. Apparently KSP is supposed to calculate conductive heat flux by part contact area. I'm not sure how that happens. Regardless, it's an issue, I'll try to work on this and see if there's a workaround or something else going on.

Examination of radiator performance.

The standard radiators, XR-500 and XR-2000, are quite well dialed in. If you ask them to dissipate the entire 550 / 2250 kW that they can draw, they stabilize at 1225 K and 1180 K, respectively. At that temperature they have a nice bright red glow, with the XR-500 starting to lean towards orange-ish.

Though perhaps their names could use adjusting, considering that the capacity numbers changed...

The conformal radiators, as it turns out, have more surface area than they need for the amount of heat that they are configured to handle.

The GR-4 is the hottest candidate, stabilizing at around 1060 K. At that point it possesses a decent, if unexciting, middle-red glow.

The GR-1 doesn't get that hot, even if dissipating 1350 kW instead of the strictly required 1250 kW. It'll stabilize around 940 K, at which point it has a dull reddish glow.

The GR-EXP, finally, is completely bored by its task and stabilizes at no more than 825 K. Its glow is so weak, it's barely visible even in the dark.

I also tested with emissiveConstant of 0.95 instead of 1.00, which generally increased the stabilization temperature by 20 K to 30 K. This didn't negatively impact any of the radiators. Even the XR-500 running at 1250 K is still in no danger of melting.

With the deployed radiators actually getting their lower conductivity after quicksaving/quickloading, the radiators also have a very generous breathing room in terms of heat transfer capacity. For example, with the "wrong" conductivity the GR-4 has so much backflow that the whole ship stabilizes somewhere a bit above 400 K, and the radiator at 225 kW transfer capacity is well balanced to keep it there. But with the correct conductivity, the radiators have no trouble cooling the entire ship to 273 K (below which they do not cool) and keep it there with performance to spare. Even adjusted to 210 kW capacity, the radiators are still easily sufficient.

Now obviously, engines also produce heat. The overcapacity of the radiators helps cool the ship during and after burns. Depends on your tastes how much overcapacity you'd like to have.

Finally, I tested just how much room for capacity the conformal radiators actually have with their areas as they are. With emissiveConstant at 0.95, I tried to get them to 1200 K, which is roughly the same heat range as the standard radiators.

The 55 kW GR-EXP can dissipate 220 kW at a temperature of 1200 K, with negligible conductive backflow

The 225 kW GR-4 can dissipate 300 kW at a temperature of 1195 K, with minor conductive backflow of about 3.5 kW

The 1350 kW GR-1 can handle 3300 kW at a temperature of 1199 K, of which 125 kW is conductive backflow and 3175 kW is radiated

I was pretty surprised by how well the GR-EXP performs, and how little difference there is to the GR-4. I mean sure, the former is meant to be mounted in sets of 4 and the latter in sets of 6, but still. Interestingly enough, the heat capacity estimation in the VAB (which I still have in my version) mirrored this from the start, showing the GR-EXP surprisingly close to the GR-4.

Good stuff. The GR-EXP probably performs well because it's technically quite large, similar to the GR-4. The main issue would be that KSP calculates part area in terms of a cube, so it's not accounting for the fact that the GR-EXP is concave on the inside, which would significantly cut into its effective radiative capabilities when open. KSP just sees it as "pretty big". The GR-4 is probably more accurately modeled in terms of effective radiating area.

I will probably adjust the names of the XR series too.

But generally, once I fix the heat conduction bug, this is pretty good, no? Awesome.

Thanks for all this testing, guys. I'm going to make you all a little... badge thing. Eventually.

[Starbuckminsterfullerton]

That's interesting how KSP uses volume instead of surface area. Are you just going to handicap the radiators that are over-estimated accordingly? As long as it performs like you want in the end and the emmissives look right and allI suppose it doesn't really matter how accurate the area is.

Propulsion seems good to go, and thermal is indeed looking pretty good, so I'd say if you feel ready you're set!

[billkerbinsky]

Noticed an anomaly with NFE X_4_11 and NFP X_5_3: built a ship out the HI-530 10m globe. hydrogen tank, VW10K quad VASIMR at the tail, and an MX-1 at the nose; four GR-1's on the MX-1, four XR2000's mounted on a battery just in front of the VW-10k. Insulators between tank and engine, and tank and reactor. Overall delta V around 110 km/s, TWR around 0.04 in high-efficiency mode.

Just after a long burn, when the reactors, engine, and radiators were hot, put the ship into timewarp, and the tank started glowing. Seems to be reproduceable -- when I warp to 1000x, the tank temperature suddenly jumped to well over 600K! When I stay at 100X or below the tank doesn't gain heat like that.

[DerekL1963]

I really should have been less sunny in my original report... I only tested a limited set of conservative configurations.

Anyhow, this is the last of the 'extreme' cases that I can think of - though 4 LV-N's attached to a TVR-400L Stack Quad-Adapter is actually a pretty common configuration. The goal here was stay clear of the red zone (temperature gauges nearly maxed out) at the end of a trans Duna injection burn. It took several iterations to get there, but the secret sauce turns out to be a FL-T400 or FL-T800 tank for thermal mass, and a quad of XR-2000's which serve to keep the main body cool and pull heat from the quad adapter and thus from the aft tanks.

Unless someone comes up with another one, I think these three (Moho, Sundiver, Quad-adapter) represent the most extreme cases the average player will usual encounter and they can all be handled easily without spamming or excessive performance degradation.

Edited May 15, 2015 by DerekL1963

[Streetwind]

But generally, once I fix the heat conduction bug, this is pretty good, no? Awesome.

Did you fix the M-EXP radiator attachment colliders yet?

*ducks and hides*

That's interesting how KSP uses volume instead of surface area.

Yeah, that also explains why the GR-1 is so powerful. Sure, it's got a huge area, but 3.3 MW potential capacity is pretty hardcore. Its wavy shape probably makes it just that much bigger in KSP's eyes.

Off the top of my head... doesn't KSP now have this huge new file where all the cubes are defined? Doesn't that imply you can specify the cube manually? Though maybe I just misinterpretated the cryptic contents of that file... well, something to investigate later, I guess. Not really ciritcal right now.

Just after a long burn, when the reactors, engine, and radiators were hot, put the ship into timewarp, and the tank started glowing. Seems to be reproduceable -- when I warp to 1000x, the tank temperature suddenly jumped to well over 600K! When I stay at 100X or below the tank doesn't gain heat like that.

Working as intended: KSP does not simulate heat at 1000x and above. At that point, the heat status of the vessel is just given the tabula rasa treatment... all equalized and frozen in time until you slow back down to x100 or below. Because of this, it doesn't matter that you insulated your tank. At that level of timewarp, KSP just doesn't care anymore.

As this is stock behavior, there's unfortunately nothing we can do about it.

Unless someone comes up with another one, I think these three (Moho, Sundiver, Quad-adapter) represent the most extreme cases the average player will usual encounter and they can all be handled easily without spamming or excessive performance degradation.

It's still silly, considering that amount of radiator power is enough to produce over 12,000 Ec/s with capacity to spare. But hey, if Squad wants the LV-N to be nonsensical like that... *shrug*

Many thanks for testing all these variations!

[void_error]

There's a chance it might happen this weekend.

Yay! Perfect. I'll have some free time on Monday if I finish my current project by then.

Keep up the good work!

- - - Updated - - -

That's interesting how KSP uses volume instead of surface area. Are you just going to handicap the radiators that are over-estimated accordingly? As long as it performs like you want in the end and the emmissives look right and allI suppose it doesn't really matter how accurate the area is.

Propulsion seems good to go, and thermal is indeed looking pretty good, so I'd say if you feel ready you're set!

Ideally, thermal behavior should also take this into account http://en.wikipedia.org/wiki/Volumetric_heat_capacity#Thermal_inertia, surface area for heat radiation ability and thermal resistance (or conductivity, same thing but backwards) for heat transfer between parts. Do the math on the launch pad and whenever you change the vessel's configuration in flight (staging, etc), store it in RAM and only update it when needed. Heavier parts like full fuel tanks will heat up slower than almost empty tanks.

[Nertea]

Okay.... new build, X.4.12. Might be the last experimental one .

Changes:

• Radiators now only show heat rejection in VAB
  
• Radiators can now be turned on or off (independant of deployed status), includes an action toggle
  
• Removed DRE's handling for radiator parts
  
• Tweaked names of universal radiators
  
• Heat pipes only transfer heat if target part is less than 1.25x the temperature of the parent part
  

[Captain Sierra]

As I have given up on fighting with the 1.0 flip-happy atmosphere (the trans-sonic drag spike is insane; I can't imagine the frustration of new players right now), I'm grabbing hyperedit and this. F*** the atmosphere, I wanna screw around with my beloved NFP. P.S. Is there an experimental version of NFP with corrected attach nodes & a quickly recompiled plugin I can run this with, or is this exclusively useful for thermal testing?

[Nertea]

Here, no promises but it's almost 100% stable. No H2 NTRs yet.

[Captain Sierra]

Here, no promises but it's almost 100% stable. No H2 NTRs yet.

Downloading this, that, and Hyperedit.

I finally got the hang of MM so I can write the H2 NTRs patches for myself (I only use stock + Atomic Age).

[billkerbinsky]

As I have given up on fighting with the 1.0 flip-happy atmosphere (the trans-sonic drag spike is insane; I can't imagine the frustration of new players right now), I'm grabbing hyperedit and this. F*** the atmosphere, I wanna screw around with my beloved NFP.

Well, if you get bored with hyperedit and want to try ground-based launches again, two tips:

- To the traditional litany of "more struts, more boosters", add "more fins.".

- If you're using NFP reactors in your deep-space stage, put them close to the nose at launch time, and put less-dense fuel tanks behind them. Managed to get a ship built around the amusingly excessive 10m-sphere LH2 tank into orbit this way.

[Streetwind]

Note that HyperEdit has some minor bugs, such as Kerbals clipping into the ground of planets they land on after their ship gets HyperEdited into orbit.

But, as it seems you're content with experimental builds, so it probably won't irk you that much.

Meanwhile, I'm grabbing X.4.12 and checking if any issues I saw before have survived the bugspray...