[0.20] Modular Fuel System 1.3/realistic fuels, reconfigurable fuel tanks and engines

[Starwaster]

You want real-world testing? You've got real-world testing. Stock install + Kerbal Engineer Redux + MFS Real Fuels, these configs.

0.1t payload:

5 tons of launcher:

LV-909: LF/OX: 7,297 m/s, LF/O2: 6,646 m/s, H2/O2: 6,767 m/s

LV-N: 3,832 m/s

50 tons of launcher:

LV-909: LF/OX: 13,062 m/s, LF/O2: 9,995 m/s, 9,665 m/s

LV-N: 10,924 m/s

...Hey, look! Pretty much everything matches up with what I said in my earlier post!

...Are you sure that you tested this?

(In hindsight, sorry if I sound snippy, but I took quite a bit of time to test this before I made my previous post, and your response was simply "you're doing it wrong")

Hey, if you did this and someone comes along and makes you feel put upon, I can understand, I mean it would be as if I'm trying to get ready for my launch window to Duna in 176 days and someone says, "Hey! Your propulsion system is broken! BROKEN!!!" (ok, a little humor...)

Unfortunately I have quite a few more mods installed. Most of them shouldn't affect this at all.

I'm using MechJeb to get my mass, TWR and deltas

I am using Arcturus' thrust corrector which changes the stock thrust output in atmo. I don't think it should affect this though, it only affects thrust and fuel consumption when in an atmosphere. (stock behavior is that engines are at full thrust output all the time regardless of isp. In atmo they pump fuel faster to compensate for the loss of thrust they should suffer)

other mods are ioncross life support, kerbal alarm clock, RCS balancer tool, KW Rocketry

with regards to config, I did this with the stock real fuel config (my own config tweak was removed for this) but I don't know that it's the same one as what you have.

If you want I'll re-do this after duplicating the conditions you tested under with regards to mods.

Right now I'm looking at:

LV909

LF/OX

wet mass 61.6

dry mass 9.3

TWR 0.08

DV: 6819 1h:03m:11s

1 KW Rocketry SB-4 LFT (3072 LF, 4608 OX)

Same ship, LVN + 6 fuel tanks

LH2

wet mass 60.5

dry mass 22.19

TWR 0.10

DV: 8377 1h:28m:19s

7 KW Rocketry SB-4 LFT (7488 LH2)

6 KW Rocketry SB-2 LFT (3744 LH2)

6 KW Rocketry SB-1 LFT (1872 LH2)

Javascript is disabled. View full album

Edit: I can't duplicate your 5 tn launcher. What are you using for command? I don't recognize that green thing, I assume that's the core

Edited August 12, 2013 by Starwaster

[bs1110101]

Would someone please make a config for tv aerospace?

[Frederf]

Have you asked the TV Aerospace creator? Usually mod authors have strong opinions on the balance of their own stuff.

[Chestburster]

Would someone please make a config for tv aerospace?

May i ask why? There are no tanks or rocket engines as far as i remember.

[bs1110101]

Then i'v found what i think is a heisenbug, because now it works!

[The Lone Wolfling]

Edit: I can't duplicate your 5 tn launcher. What are you using for command? I don't recognize that green thing, I assume that's the core

KW rocketry's fuel tanks have a higher mass ratio than stock parts, which is why they end up with those results.

As I said, this is with the stock game + realfuels + kerbal engineer redux.

The "command pod" you see is the Kerbal Build Engineer surface-mounted on a Probodobodyne OKTO. Total mass 0.1t.

[Starwaster]

KW rocketry's fuel tanks have a higher mass ratio than stock parts, which is why they end up with those results.

As I said, this is with the stock game + realfuels + kerbal engineer redux.

The "command pod" you see is the Kerbal Build Engineer surface-mounted on a Probodobodyne OKTO. Total mass 0.1t.

I have to concede the point that there is a problem here. I'm not sure I agree on where it is but it's definitely there, and I'm sorry I said that you hadn't tested or that you were doing it wrong. I've spent the better part of the morning (too long in fact, have to get busy) going over your past posts and the extent to which you've tried to solve this is pretty damned obvious.

I'm not clear why the KW parts would have a higher mass ratio since their mass is calculated the same as everything else's, based on their volume, but I'll take your word for it.

I think where I went wrong is that I didn't go far enough in testing, and the problem manifests itself worse and worse as you go on. I guess the denser fuels either aren't hitting that problem or aren't hitting it until much farther down the line, farther than any any rocket that anyone would practically try to build.

So, what I think is that the NERVA should have some of its stats bumped up. The stock version basically represents an actual prototype dating back to early 70s, I'm sure a production model would have performed better (not trying to tell you something I'm sure that you already know). ISP of 1000 would be attainable by a NERVA type rocket. Some of them also had thrust levels of 333 kN or higher.

The other thing that I noticed that you (and others) have focused on in trying to solve this is tank mass and the first thing I see looking at tank masses that I don't see actually talked about in this thread is the fact that tank base masses are based on volume. This is wrong. Tank mass should be based on surface area and surface area to volume ratio drops as volume goes up. LH2 tanks wouldn't be penalized nearly as badly as they are now where tank mass increases linearly with volume.

I'm not sure though how you would calculate something like surface area on volume alone though.

[The Lone Wolfling]

I have to concede the point that there is a problem here. I'm not sure I agree on where it is but it's definitely there, and I'm sorry I said that you hadn't tested or that you were doing it wrong. I've spent the better part of the morning (too long in fact, have to get busy) going over your past posts and the extent to which you've tried to solve this is pretty damned obvious.

I'm not sure how to respond to someone the internet conceding a point. ...Thank you.

I'm not clear why the KW parts would have a higher mass ratio since their mass is calculated the same as everything else's, based on their volume, but I'll take your word for it.

Look at the tank mass ratios of the album posted earlier versus stock fuel tanks.

I think where I went wrong is that I didn't go far enough in testing, and the problem manifests itself worse and worse as you go on. I guess the denser fuels either aren't hitting that problem or aren't hitting it until much farther down the line, farther than any any rocket that anyone would practically try to build.

What's happening is that the rockets with higher specific

So, what I think is that the NERVA should have some of its stats bumped up. The stock version basically represents an actual prototype dating back to early 70s, I'm sure a production model would have performed better (not trying to tell you something I'm sure that you already know). ISP of 1000 would be attainable by a NERVA type rocket. Some of them also had thrust levels of 333 kN or higher.

I've tried that. See here. The problem is that it breaks the game - the NERVA and advanced fuels in general ends up being too good.

The other thing that I noticed that you (and others) have focused on in trying to solve this is tank mass and the first thing I see looking at tank masses that I don't see actually talked about in this thread is the fact that tank base masses are based on volume. This is wrong. Tank mass should be based on surface area and surface area to volume ratio drops as volume goes up. LH2 tanks wouldn't be penalized nearly as badly as they are now where tank mass increases linearly with volume.

I'm not sure though how you would calculate something like surface area on volume alone though.

Radius is proportional to the cube root of volume. Surface area is proportional to the square of the radius. Therefor, surface area is proportional to volume to the 2/3 power.

Although this assumes spherical tanks. In actuality, it would be proportional to the volume, but with a different constant for the different radii of tank.

If you think about it, assuming a cylindrical tank of constant radius, the surface area and hence mass is proportional to volume.

Finally, I talked about this briefly, but discarded it as it isn't really practical in KSP, as we cannot currently say "I want a 2.4m radius X 4.9m height tank". If/when we get procedural fuel tanks, though...

[Fractal_UK]

So, what I think is that the NERVA should have some of its stats bumped up. The stock version basically represents an actual prototype dating back to early 70s, I'm sure a production model would have performed better (not trying to tell you something I'm sure that you already know). ISP of 1000 would be attainable by a NERVA type rocket. Some of them also had thrust levels of 333 kN or higher..

If you look at the thrust and specific impulse of the real NERVA it's easy to come to the conclusion that the stock LV-N is underpowered but the moment you account for the mass of the nuclear reactor, you will realise that the opposite is in fact the case. The NERVA prototype massed at 34 tons to provide those 334kN of thrust, the stock model weighs only 2.25 tons but provides 60kN. The power to weight ratio of nuclear reactors moves upward fairly quickly with increasing mass, so its totally ridiculous that such a tiny reactor should have such a huge power output.

This is a problem I've grappled with in my own mod (where the largest nuclear reactor has almost identical stats to the real NERVA). I'm trying to solve the problem by allowing the nuclear thermal rockets to switch between a variety of propellants on the fly. There is no reason that they have to use LH2, it is the most efficient propellant, of course, but it is hard to store and requires huge tankage. For that reason, other propellants might work better in practice.

[Starwaster]

I'm not sure how to respond to someone the internet conceding a point. ...Thank you.

It happens sometimes. Strange, but true.

I've tried that. See here. The problem is that it breaks the game - the NERVA and advanced fuels in general ends up being too good.

Well I could always do my Duna mission with the LV909s... but that's just wrong somehow.

Radius is proportional to the cube root of volume. Surface area is proportional to the square of the radius. Therefor, surface area is proportional to volume to the 2/3 power.

Although this assumes spherical tanks. In actuality, it would be proportional to the volume, but with a different constant for the different radii of tank.

If you think about it, assuming a cylindrical tank of constant radius, the surface area and hence mass is proportional to volume.

Finally, I talked about this briefly, but discarded it as it isn't really practical in KSP, as we cannot currently say "I want a 2.4m radius X 4.9m height tank". If/when we get procedural fuel tanks, though...

What you're saying also means that radius has to increase a lot more before my stated concerns become valid anyway. And our volume isn't even dependent on those factors anyway, we get it arbitrarily assigned from config files. But I think for the most part, those values do seem to be accurate if we make certain assumptions about what those values represent (i.e. 'kerbos'). So I think what's really bothering me about this is that we're using some arbitrary fictional unit of measurement (that is equal to about 160 cubic meters) and multiplying it by a known mass unit of measurement. I think it ought to be looked again.

[Cashen]

Not sure if I totally buy the idea of KW's tanks being dramatically different from the stock ones. The metric I used for the performance of a bipropellant tank was "liters of fuel per ton of mass" and most stock tanks are around 79-80L per ton regardless of the total volume, and KW is around this as well, maybe as high as 81 in some cases, but a very small difference. NovaPunch also follows similar guidelines, though their largest 1.25m tank does hold around 84L per tonne, the biggest deviation I found.

[Starwaster]

Not sure if I totally buy the idea of KW's tanks being dramatically different from the stock ones. The metric I used for the performance of a bipropellant tank was "liters of fuel per ton of mass" and most stock tanks are around 79-80L per ton regardless of the total volume, and KW is around this as well, maybe as high as 81 in some cases, but a very small difference. NovaPunch also follows similar guidelines, though their largest 1.25m tank does hold around 84L per tonne, the biggest deviation I found.

Not liters... do you mean kiloliters perhaps? I'm not sure what sense to make of this...

In any case, all fuel tanks affected by the mod have their part mass replaced by a basemass of (usually) 0.000125 kg * volume + another fuel specific tank mass for every different fuel the tank carries. (configurable per tank per fuel type)

Edit: Except tanks that don't HAVE basemass specified... then they do look at the part file. (sorry, reading this from the source to get a better grip on how the mod is handling fuel tank mass)

Edited August 13, 2013 by Starwaster

[The Lone Wolfling]

Not liters... do you mean kiloliters perhaps? I'm not sure what sense to make of this...

In any case, all fuel tanks affected by the mod have their part mass replaced by a basemass of (usually) 0.000125 kg * volume + another fuel specific tank mass for every different fuel the tank carries. (configurable per tank per fuel type)

Really? Look at your imgur album.

Because the mass ratios of the tanks shown don't match with the tank mass ratios of stock tanks. In particular the mass ratio of LF/OX is lower. (Sorry, I seem to have gotten that wrong earlier. The other mass ratios are above stock, but this one is lower. Still the same effect though.) this screenshot. Tank mass ratio is 57.01 / 4.799 = 11.88. Compare that with the stock orange tank at 44.47 / 0.96 = 46.32.

This shows that that formula is incorrect.

[Starwaster]

Really? Look at your imgur album.

Because the mass ratios of the tanks shown don't match with the tank mass ratios of stock tanks. In particular the mass ratio of LF/OX is lower. (Sorry, I seem to have gotten that wrong earlier. The other mass ratios are above stock, but this one is lower. Still the same effect though.) this screenshot. Tank mass ratio is 57.01 / 4.799 = 11.88. Compare that with the stock orange tank at 44.47 / 0.96 = 46.32.

This shows that that formula is incorrect.

I think that's because the tank types are set different

The big orange tank's fuel type is set to Cryogenics. The KW is set to Default.

@PART[fuelTank3-2]
{
    MODULE
    {
	name = ModuleFuelTanks
	volume = 6400
	type = Cryogenic
    }
}

@PART[KW2mtankL4]
{
    MODULE
    {
		name = ModuleFuelTanks
		volume = 7680
		type = Default
    }
}

Then if you look in RealTankTypes.cfg

Cryo basemass = 0.00015

Default basemass = 0.000125

Edit: Was going to do a lengthier post but I had to go watch 'Under the Dome' to watch the further adventures of Big Jim Rennie.

Here's something interesting about the Cryogenics tank type: The orange tank (more precisely the Cryogenics tank type) is not assessing any extra tank overhead for LiquidFuel or Oxidizer. But the config file specifies extra mass for those fuel types. Not sure why but seems like basically the orange tank lets you use LF/OX at a mass discount. Tank should definitely be a bit more massive than 0.96 tons.

Edit 2: This seemed awfully familiar but I couldn't place it. Had to go back a few pages. Apparently the realfuel tank config file is trying to modify non-existent lines when it should be inserting new lines. That tank should be at least 3 tons more massive when carrying liquid fuel & oxidizers. And it's netting Big Orange a bonus delta-v of 1357 m/s

Edited August 13, 2013 by Starwaster

[The Lone Wolfling]

Edit 2: This seemed awfully familiar but I couldn't place it. Had to go back a few pages. Apparently the realfuel tank config file is trying to modify non-existent lines when it should be inserting new lines. That tank should be at least 3 tons more massive when carrying liquid fuel & oxidizers. And it's netting Big Orange a bonus delta-v of 1357 m/s

Hmm...

Sounds like that might be the cause of the issues.

[Starwaster]

Hmm...

Sounds like that might be the cause of the issues.

I did this as a patch file instead of editing the originals, otherwise this would have been a change to realTankTypes.cfg

@TANK[Cryogenic]
{
	@TANK[LiquidFuel]
	{
		mass = 0.0005
	}
	@TANK[Oxidizer]
	{
		mass = 0.0005
	}
}

1 three man command module + 1 big orange tank + 1 LV909

(LF/OX)

mass (wet/dry)

52.29 / 8.66

DV 6472 (52min 38s)

Brings it more in line with what the results that you saw with the KW (I think that was getting ~6800ish DV as that tank has a bit more capacity)

(didn't check the mass ratio, busy busy busy)

LF/LO2 might get a little love now...

Delta-V 6664

LO2/LH2 meh

Delta-V 5296

Can't do the LVN because it's modded right now

Edit: Ooops LH2 and also doesn't have any extra tank mass in that tank. Not sure if that's intentional; maybe because the cryo tank already has some extra overhead factored in.... but you'd think it would at least match LO2 as it requires even more refrigeration. LO2 has tank mass: 0.000475

Edited August 13, 2013 by Starwaster
Oops

[Starwaster]

Small 'patch' file for the above mentioned LF/OX tank mass issue. (only if you're using the 'Real Fuels' option)

I've gone through everything else and didn't see anything glaringly obvious. I think LH2 should probably have some additional tank mass in the Cryogenic tank type but it's explicitly set to 0.0 so I was reluctant to touch it as I don't know what ialdabaoth's intention was.

I did notice that the Fuselage tank type was also missing any tank fuel mass entries or basemass values. So it's basically just using a static dry mass based on the part mass.

RealFuel_MissingTankMass_Fix.cfg

Just put it somewhere in your modularfueltanks directory. Or actually I think it can even go anywhere in your GameData directory...

Edited August 13, 2013 by Starwaster

[bs1110101]

Now we need configs for the new engines in the Kosmos pack.

[RuBisCO]

The only way water is going to have performance 'about the same' as LO2/LH2 is if you break it up into and then liquefy it as LO2 and LH2. You've just made rocket fuel! That's not the same thing as saying that you are using water as a propellant. But you *CAN* use water as a propellant directly in a nuclear drive; it would have an ISP of approximately 412.

The ISP is dependent on the engines outlet temperature or in this case the reactor's. A reactor outlet temp of 2520 K comes out to an ISP of 412, but reactor temp of ~3500 K should be possible, this is beyond the temp of LH2+LO2 combustion (2,985 K) and should be a higher ISP.

[Starwaster]

The ISP is dependent on the engines outlet temperature or in this case the reactor's. A reactor outlet temp of 2520 K comes out to an ISP of 412, but reactor temp of ~3500 K should be possible, this is beyond the temp of LH2+LO2 combustion (2,985 K) and should be a higher ISP.

Actually the ISP I listed was assuming a reactor temperature of 3200k (solid core, NERVA style).

If you have some data that says you can get a higher ISP with water than 400 I'm interested in seeing it but compared to other articles I've found it's a pretty generous rating. Mostly everything I've seen on the subject of using water as a propellant in a nuclear rocket cited an ISP of no higher than 200. It's just not a very efficient propellant at all. It's something you'd use in a pinch, say if you were landed on Mars and the mission plan required you to harvest your own fuels from the environment. Something you'd use because you COULD Not because it were desirable.

Honestly I'm not sure how to go about calculating it from scratch myself but I trust the 412 number because I trust the source. Project Rho. There's a wealth of rocketry related information there and I find it to be generally reliable. Sometimes annoyingly so

[RuBisCO]

Actually the ISP I listed was assuming a reactor temperature of 3200k (solid core, NERVA style).

If you have some data that says you can get a higher ISP with water than 400 I'm interested in seeing

You can get any ISP from water it just a matter of how hot you heat it to. This is a matter of logic though: the exhaust of LH2+LO2 is water (assuming stoichiometry optimal combustion ratio, now if it hydrogen rich that would add ISP up until it cools the combustion temperature too much) its ISP maxes out IRL engines at roughly ~460 with a combustion temp of roughly ~3000 k (see http://www.astronautix.com/props/loxlh2.htm and sublinks on engines) a nuclear thermal rocket should achieve the same if operating at the same temp.

Back to math: if we calculate ISP via the equations

ISP=Ve/g , Ve=SQRT((2*k/(k-1))*(R*Tc/M)*(1-(Pe/Pc)*(k-1)/k))

k= ~1.25 +/- .08

R= 8314.51 N-m/kg-K

Tc= 3500 K

M = 18 amu

Pe = 1 atm

Pc = 68 atm

The question becomes what is k? Steam's specific heat ratio (k) is in flux because the hotter steam gets and the low its pressure the lower its ratio gets. Steam is far from an ideal gas (especially when it hot enough to have a significant percentage of it dissociating into oxygen and hydrogen) and its k is difficult to calculate and best based on real measurements for example here although are far too low for the temps we are looking at. Here on this site someone ask the very useful question of how to calculate ISP for nuclear thermal rocket and someone posts an uncited table of k for water (steam): at 3000 K we get a value for k of 1.175 which would compute to an ISP of 439 at 3000 K. At 3500 K, k would be ~1.171 and ISP would be 479.

it but compared to other articles I've found it's a pretty generous rating. Mostly everything I've seen on the subject of using water as a propellant in a nuclear rocket cited an ISP of no higher than 200.

I would like to see a citation for that as well, especially its outlet temp.

It's just not a very efficient propellant at all.

There more than efficiency in ISP here. Water is readily minable in the solar system and does not require cryogenic storage. In this proposal here using even a low ISP nuclear thermal rocket derived directly from today's reactor technology (not a theoretical nuclear engine design) with a 1100 K exhaust temp, 198 ISP, to launch to lunar orbit water that is mined and melted from lunar permafrost would require ~1/10 to ~1/100 the mass of equipment to the moon verse minning, melting, cracking, cooling and storing LO2 and LH2 and getting equal amount of fuel to orbit per year.

If we were to have in KSP minning other then magical 'kethane', water would be it, and if we were to have that level of realism in KSP then nuclear thermal rockets using water should be an option as well.

[Starwaster]

You can get any ISP from water it just a matter of how hot you heat it to. This is a matter of logic though: the exhaust of LH2+LO2 is water (assuming stoichiometry optimal combustion ratio, now if it hydrogen rich that would add ISP up until it cools the combustion temperature too much) its ISP maxes out IRL engines at roughly ~460 with a combustion temp of roughly ~3000 k (see http://www.astronautix.com/props/loxlh2.htm and sublinks on engines) a nuclear thermal rocket should achieve the same if operating at the same temp.

Back to math: if we calculate ISP via the equations

ISP=Ve/g , Ve=SQRT((2*k/(k-1))*(R*Tc/M)*(1-(Pe/Pc)*(k-1)/k))

k= ~1.25 +/- .08

R= 8314.51 N-m/kg-K

Tc= 3500 K

M = 18 amu

Pe = 1 atm

Pc = 68 atm

The question becomes what is k? Steam's specific heat ratio (k) is in flux because the hotter steam gets and the low its pressure the lower its ratio gets. Steam is far from an ideal gas (especially when it hot enough to have a significant percentage of it dissociating into oxygen and hydrogen) and its k is difficult to calculate and best based on real measurements for example here although are far too low for the temps we are looking at. Here on this site someone ask the very useful question of how to calculate ISP for nuclear thermal rocket and someone posts an uncited table of k for water (steam): at 3000 K we get a value for k of 1.175 which would compute to an ISP of 439 at 3000 K. At 3500 K, k would be ~1.171 and ISP would be 479.

I would like to see a citation for that as well, especially its outlet temp.

There more than efficiency in ISP here. Water is readily minable in the solar system and does not require cryogenic storage. In this proposal here using even a low ISP nuclear thermal rocket derived directly from today's reactor technology (not a theoretical nuclear engine design) with a 1100 K exhaust temp, 198 ISP, to launch to lunar orbit water that is mined and melted from lunar permafrost would require ~1/10 to ~1/100 the mass of equipment to the moon verse minning, melting, cracking, cooling and storing LO2 and LH2 and getting equal amount of fuel to orbit per year.

If we were to have in KSP minning other then magical 'kethane', water would be it, and if we were to have that level of realism in KSP then nuclear thermal rockets using water should be an option as well.

I think your last paragraph hits on it perfectly. Water as a propellant is really no more than a curiousity in KSP unless it actually becomes part of the game to the point that you can mine water (or ice) and actually use it as a resource for your ships. Also necessary things for it to become meaningful is that we need to have costs and budgets implemented. Then, when you're low on funds, your deep space mission is struggling and then it starts to get meaningful. But one thing that has to happen (game standpoint) is that we need to be able to reconfigure engines outside of the VAB. The NERVA has the advantage that you can realistically pump a wide variety of things in there that you couldn't do with a chemically fueled rocket. (at least I don't *THINK* you can just chuck water in there... that's not going to work)

I've got a config that I'm playing with that I can share where I've already implemented quite a few propellants. Water's in there but the rocket isn't set up to allow it to be configured yet. (not hard to do, just haven't gotten to it). In fact I pasted it in a message a few pages back but that one is a bit broken.

I saw the neofuel site, that's one of the sources I was investigating; that's where I saw the lower values in the 200 range for isp but I thought it was for a hotter reactor. I was reading up on it again and they're assuming a reactor of only 800K. I'm not sure why unless there's a performance hit for heating it up too much, which you alluded to above. Which runs counter to everything else I've read on the subject, which says you want to heat it up (your nuclear propellants) to the point that it disassociates into its lowest molecular weight gasses. (so, not even steam at that point).

[RuBisCO]

I been able to set up nuclear engines to operate on different fuels in VAB, but once set they can't switch in flight.

IRL yes a nuclear thermal rocket engine can be designed to operate on a variety of fuels, but probably could not switch post manufacturing. For example H2O and H2 have very different properties such that being able to switch in flight would present all sorts of problems for pumping, the fuel element coatings, etc, etc. CH4 and CO2 present problems of carbon soot build up and then there is the problem of free oxygen and highly corrosive environment that fuels like CO2 and H2O could present, H2 present problems of metal brittleness and reducing environment. Fuel rods would need to be coated with or made of reduced compounds (metals) for H2, NH3 and CH4 but would need to be made of or ceramic oxides for CO2 and H2O. A pebble/dust bed centrifuged reactors could get over some of these problems by removing structural strength of fuel elements as a concern: as the fuel is then held in place by centrifugation rather then mechanical strength making opening up a larger variety of fuel element types and coatings that don't need to be strong and could operate with a wide variety of fuels.

I would also like to go back to the point of hydrogen richness of LO2/LH2 engines. The modern Centaur RL10 engine operates on 20.3 kg LO2 and 3.5 kg LH2 per second (according to wiki). That comes to a stoichiometric ratio of 2.72 mol of H2 per mol of O2, ideally the ratio would be 2 for 100% combustion, so the left over H2 makes the combustions H2 rich, in total assuming 100% combustion only 4% of the exhaust by mass is the rich H2, with a outlet temp estimated at 3000 K that 4% has an ISP of 946, the rest has an ISP of 439: total 459. That is pretty close to RL10 measured vacuum ISP of 464. In short the realistic amount of ISP improvement from running hydrogen rich verse pure water output is ~20 s. So a rule of thumb we could go by is that a nuclear rocket operating at the same exhaust temp as LO2/LH2 (~3000 K) rocket should be 20 ISP seconds less then the LO2/LH2 rocket. We would need to bring up the temp by 300 K to make up the difference. 3500-3800 K is probably be best we could expect from even a drum reactor without going with a liquid core drum reactor: that would come to max theoretical ISP of 487 and 1050 for Water and LH2. A liquid core (drum) reactor is claimed to be able to get 1350-1500 ISP with hydrogen which would be a outlet temp of ~7500 K which I don't believe because even tungsten boils at 5800 K. A Gas Core though operating at 25,000 K with an ISP by my rough calculation (assuming nearly complete dissociation into protons and oxygen) of ~3900 for LH2 and ~1660 for H2O.

I would like to see these options beyond just a NERVA facsimile in KSP

(centrifugal) particle core reactor with ans ISP of 1050 using LH2 and a thrust to weight ratio twice that of the ingame NERVA

(centrifugal) liquid core reactor with an ISP of 1450 using LH2 and a thrust to weight ratio similar to the ingame NERVA

Gas Core reactor with an ISP of 3800 using LH2 and a thrust to weight ratio of say 1/4 that of NERVA and rapid use of nuclear fuel as well as some kind of means to prevent us on or near Kerbin because of all the nuclear waste it would spit out.

I guess all of these could be on a tech tree from NERVA solid core reactor and up.

Of course my favourite nuclear engine is nuclear salt-water rocket a continuous nuclear explosion engine that make Orion look stupid! Using dissolve liquidized nuclear fuel in water and making it go super-critical in a rocket chamber, adding alot more water to cool the chamber and as propellent mass average exhaust temps of ~200,000 K an ISP of over 6000 is possible (and that is using water!) Such and engine would also have incredible thrust. Of course ingame such an engine would need to be incredible expensive, big and prohibited from operating anywhere near kerbin or any other planet that could be habited without a space suit (and radiation shielded suits) because this engine spews radioactive waste.

Edited August 14, 2013 by RuBisCO

[Starwaster]

I been able to set up nuclear engines to operate on different fuels in VAB, but once set they can't switch in flight.

IRL yes a nuclear thermal rocket engine can be designed to operate on a variety of fuels, but probably could not switch post manufacturing. For example H2O and H2 have very different properties such that being able to switch in flight would present all sorts of problems for pumping, the fuel element coatings, etc, etc. CH4 and CO2 present problems of carbon soot build up and then there is the problem of free oxygen and highly corrosive environment that fuels like CO2 and H2O could present, H2 present problems of metal brittleness and reducing environment. Fuel rods would need to be coated with or made of reduced compounds (metals) for H2, NH3 and CH4 but would need to be made of or ceramic oxides for CO2 and H2O. A pebble/dust bed centrifuged reactors could get over some of these problems by removing structural strength of fuel elements as a concern: as the fuel is then held in place by centrifugation rather then mechanical strength making opening up a larger variety of fuel element types and coatings that don't need to be strong and could operate with a wide variety of fuels.

That's creating a new part or redefining the existing nuclearEngine to be something else which I wanted to stay away from.

I understand about the need for different linings but to be quite frank I don't think that would be as much fun. Assuming as before that we end up with a game that allows for harvesting fuel from the environment in-situ. (I think we'll have to get there via mods though, I'm not sure I see the final KSP product going quite that far with resources, if we actually do get them)

Edit: There is a way to switch in-flight: Hybrid engines. I haven't been able to make it work, I messed something up and the rocket ate all its nuclear fuel instantly on ignition. But it does let you switch between two different types. (exactly two, you can't specify multiples ) And I'm not sure you can actually configure it in the VAB.... (that is, it has to be set up in the part's CFG or via MM patch)

Edited August 14, 2013 by Starwaster

[RuBisCO]

That's creating a new part or redefining the existing nuclearEngine to be something else which I wanted to stay away from.

Well people do make new engines parts, the "Fat Man" series for example.

I understand about the need for different linings but to be quite frank I don't think that would be as much fun.

I was saying IRL, what done ingame is different. But as a matter of coding is seems easier just to program the engines fuel type in the VAB and have it locked as such from launch onward. I've been able to do this using the Realfuels as a guide and making Ammonia and Water as extra fuel options for Nuclear engines. I've also been playing around with the Kethane mod to make it Water and make only H2,O2,H2O2 (and of course Water)