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

[xZise]

But what could be wrong?! Simply saying "THERE IS SOMETHING WRONG" doesn't help. With an Isp of 3000 the LV-N would be almost as efficient as the Xenon Ion Engine. Or do you also want to pimp the Isp of that engine to 42000? The only reasonable way to “maintain realism†is to reduce the mass. This would increase the ÃŽâ€v as you could add more fuel on your craft. You could for testing purposes change the mass of the LV-N to the mass of the LV-T45 and then check the results?

What I'm trying to say is: Physically speaking it works fine! So I don't know where the problem might be that it (seems?) unrealistic, but there is another problem about realism: There was never a nuclear engine mounted on a rocket so we don't know how much fuel it would need. The LV-N needs not very much mass but a lot of volume and this adds additional dry mass making your over all ÃŽâ€v worse.

About the Space Shuttle example: The SSMEs (and the LV-T45 when fitted to LH2+LOX) are burning both and as LOX is very dense (compared to LH2) a SSME/LV-T45 burns longer with the same volume of fuel.

Another way to test that would be to test the LV-N with a heavier craft, so that the additional mass by the LV-N doesn't matter.

Okay I tried to rebuild rosenkranz's ships but my LF/OX craft (build like yours) was able to add one FL-T100 and one Oscar-B fuel tank and the mass is only 5.48 t? Now the dry mass for my ship is then 2.07 t so there are 3.43 t fuel on board. With a LV-N rocket like you build and reduced the mass of the LV-N to 1.5 (LV-T45's) I was able to put 2.45 t of fuel in it. That explains your discrepancies: LH2 (and LOX) drymass/fuelmass ratio is worse than the LF/OX ratios. Now I don't have a good solution for that, except maybe to either use large amounts of LH2 (at least you choose to do the hardcore mode) or stick to classic engines which use the heavier LF, OX and LOX.

By the way: With the reduced mass of the LV-N I get 4.9 km/s ÃŽâ€v (due to the additional fuel I was able to add).

Fabian

Edited August 2, 2013 by xZise

[AngelLestat]

But what could be wrong?! Simply saying "THERE IS SOMETHING WRONG" doesn't help. With an Isp of 3000 the LV-N would be almost as efficient as the Xenon Ion Engine. Or do you also want to pimp the Isp of that engine to 42000? The only reasonable way to “maintain realism†is to reduce the mass. This would increase the ÃŽâ€v as you could add more fuel on your craft. You could for testing purposes change the mass of the LV-N to the mass of the LV-T45 and then check the results?

But you really did some testing with nerva engines using this mod???????

And I mean with the 0.21 version

You saw the images that I post in the example??

how can you say you do not find anything unusual about that???

Is a 6T engines with 300 of thrust and 900 ISP, then 13000 of H2 and one little capsule. And I get like 2000 of delta v in vaccum. BUT.... you still cant find nothing wrong there... haha

And when I mention about remplace the isp value of the nerva engines with a crazy number, is a simple way to fix something that is not working how it should. I know that the nerva isp value is not the problem.. The problem is somewhere in how the game engine calculates thrust using only H2. AND THERE IS THE PROBLEM.

There was never a nuclear engine mounted on a rocket so we don't know how much fuel it would need. The LV-N needs not very much mass but a lot of volume and this adds additional dry mass making your over all ÃŽâ€v worse.

Becoz we never need carry big payloads to mars or jupiter. And becoz is NUCLEAR, so it has its risk. If the rocket explodes in our atmosphere we will have some eco environment problems.

Another way to test that would be to test the LV-N with a heavier craft, so that the additional mass by the LV-N doesn't matter.

This is the proof that you dint see my previous examples.

[rosenkranz]

Even following and participating in this conversation and I'm no longer sure what we're arguing about...

I was agreeing that something was wrong. In theory going with the 'better' fuels should yield benefits which my testing was not getting. (note, on those craft, I had to remove some fuel to get the mass exactly the same. I used the tanks that would require the least amount of fuel to be removed)

One of the conclusions that another poster made was that it was the 'out of whack' dry masses of the tanks that was the cause.

The NERVA argument was another matter. This mod is also bringing to light the potential fact that the NERVA isn't quite all that an a bag of chips like we have all come to expect. NERVA's arn't meant to be launch engines as it has a TWR that is way to weak to put something in orbit. You'll lose too much to gravity no matter how much tankage you have.

The NERVA stages I made were hypothetical upper stages that would start from orbit. What is killing it is the tankage's dry mass ratio to the fuel within. I'd try playing with the dry mass of the tanks just to test this but that is a bit more science than I really care to do. But I suspect NERVA's under this mod won't do well without special LH2 tanks.

I'm thinking deployable tanks. Tanks the have fuel loaded at launch have got to be able to stand up under their own weight on Kerbin. Thus I wouldn't really recomend changing existing tanks.

As I type this I'm wondering if we could maybe get Hooligan labs to develop a deployable envelope that could hold fuel........ One could hope he's listening

Anyway, I think NERVA ships would have to be built in orbit rather than launched (or launched with undelployed tanks)

[The Lone Wolfling]

But what could be wrong?! Simply saying "THERE IS SOMETHING WRONG" doesn't help.

Alright.

So, the problem with the LV-N (and LOX/LH2 in general) is that the tank mass ratio is too low for it to make sense to use it.

The only reasonable way to “maintain realism†is to reduce the mass. This would increase the ÃŽâ€v as you could add more fuel on your craft.

Yes it would increase the delta v, but even if the mass of the NERVA was the same as the LV-45, it still isn't worth it.

What I'm trying to say is: Physically speaking it works fine!

Wrong. IRL mass ratios are much higher.

[...] but there is another problem about realism: There was never a nuclear engine mounted on a rocket so we don't know how much fuel it would need.

Wrong again. We know how much fuel a NERVA would need as we've conducted tests. Not to mention that it's pretty easy to calculate approximate exhaust velocities of a purely thermal rocket, all things considered.

The LV-N needs not very much mass but a lot of volume and this adds additional dry mass making your over all ÃŽâ€v worse.

Correct in KSP. IRL, again, the mass ratios are higher and as such the LV-N does make sense in delta-v terms.

Another way to test that would be to test the LV-N with a heavier craft, so that the additional mass by the LV-N doesn't matter.

...A couple posts back I did the math for an infinite mass ship, and it still doesn't make sense to use a LV-N.

Something needs to be done about the LV-N. Currently with the real fuels it is never worth it to use a LV-N.

[pina_coladas]

Yeah guys this is all about the fuel tank mass ratios that are much worse in ksp than in real life. Squad has balanced things in a slapdash way with only one set of fuels in mind, and they will face the exact same rebalancing problems if and when they add more fuel types in the future.

This thread is fun because it highlights (in an exaggerated way) the drawbacks of nuclear thermal rockets that eat away at the dv gain you get from the high ISP; you have to consider mass ratios and ISP together, and the lowest density of all fuels plus the high mass of the engines hurt your mass ratio. KSP masses are apparently weird enough that the worse mass ratio of NERVAS completely negate the increased ISP!

So yeah this mod will have to be totally rebalanced to make sense in the KSP universe. Maybe all fuel tank dry masses can be lowered by the plugin? It would be a pretty dramatic change to the balance of the game but maybe it's worth it.

[NathanKell]

@ialdabaoth, keep your priorities right-side up. Deal with RL first. We'll keep.

[The Lone Wolfling]

Yeah guys this is all about the fuel tank mass ratios that are much worse in ksp than in real life. Squad has balanced things in a slapdash way with only one set of fuels in mind, and they will face the exact same rebalancing problems if and when they add more fuel types in the future.

Yep.

This thread is fun because it highlights (in an exaggerated way) the drawbacks of nuclear thermal rockets that eat away at the dv gain you get from the high ISP; you have to consider mass ratios and ISP together, and the lowest density of all fuels plus the high mass of the engines hurt your mass ratio. KSP masses are apparently weird enough that the worse mass ratio of NERVAS completely negate the increased ISP!

Yep again.

So yeah this mod will have to be totally rebalanced to make sense in the KSP universe. Maybe all fuel tank dry masses can be lowered by the plugin? It would be a pretty dramatic change to the balance of the game but maybe it's worth it.

I've tried doing that locally, and it breaks the game, or rather it makes it much much much easier to get anywhere to the point where you don't need the NERVAs anyways.

You hit the decreased size of the planets, in other words.

[pina_coladas]

I've tried doing that locally, and it breaks the game, or rather it makes it much much much easier to get anywhere to the point where you don't need the NERVAs anyways.

You hit the decreased size of the planets, in other words.

Maybe the ISP advantage from low density fuels needs to be exaggerated a bit instead, just enough that you actually come out ahead. Would this be any better than messing with fuel tank masses? Either way, it's clear that realistic values do not work well in the ksp universe with its wacky scale.

[NathanKell]

One potential fix? Have MFT change the mass of the tank when you switch fuel mode. Does LH2 have lower pressure on tank walls (for its rated density) than LF/Ox? If so, don't need as strong a tank, so lower the mass of the tank dynamically.

[The Lone Wolfling]

Maybe the ISP advantage from low density fuels needs to be exaggerated a bit instead, just enough that you actually come out ahead. Would this be any better than messing with fuel tank masses? Either way, it's clear that realistic values do not work well in the ksp universe with its wacky scale.

Hmm...

It would certainly have the advantage of not changing stock equivalents at all.

Let's see. The LV-909 has a specific impulse in LF/OX mode of 370. It would have to have a LF/LOX Isp of at least 526(!) to be worth it, and a H2/LOX Isp of at least 648(!).

The NERVA would have to have an Isp of >1030(!!) to be worth it.

It would be doable, as long as people didn't complain about the unrealistic values. Want me to whip up a temporary set of config files to test?

One potential fix? Have MFT change the mass of the tank when you switch fuel mode. Does LH2 have lower pressure on tank walls (for its rated density) than LF/Ox? If so, don't need as strong a tank, so lower the mass of the tank dynamically.

Thought about this, but the dry tank mass would have to be so low...

Not to mention that cryogenic fuels tend to require more tank mass, due to the amount of insulation required.

Edited August 3, 2013 by The Lone Wolfling

[AngelLestat]

rosenkranz

This mod is also bringing to light the potential fact that the NERVA isn't quite all that an a bag of chips like we have all come to expect. NERVA's arn't meant to be launch engines as it has a TWR that is way to weak to put something in orbit. You'll lose too much to gravity no matter how much tankage you have.

I did many tests in orbit and the delta V that I get is still damm low. So is not a gravity problem.

The Lone Wolfling

So, the problem with the LV-N (and LOX/LH2 in general) is that the tank mass ratio is too low for it to make sense to use it.

I am not agree.

Yes.. the tank mass ratio can in fact have something to do. But the huge difference that we saw in performance it need to have another cause.

Something tells me that KSP is converting only the half of the fuel into thrust when it use only H2

We have the normal fuel ratio of 0.9 and 1.1 for oxidexer that the game usual use. So... Is possible that when the mod tells to ksp that it has only H2 to burm, then it use 0,9 of H2 and 1.1 extra of H2 to get the same thrust? So it will burn 2 times the H2 needed?

That will explain why we get less than 1/4 of the delta V that we get in the same weight example without the mod.

Edited August 3, 2013 by AngelLestat

[pina_coladas]

Hmm...

It would certainly have the advantage of not changing stock equivalents at all.

Let's see. The LV-909 has a specific impulse in LF/OX mode of 370. It would have to have a LF/LOX Isp of at least 526(!) to be worth it, and a H2/LOX Isp of at least 648(!).

The NERVA would have to have an Isp of >1030(!!) to be worth it.

Well, do the stock ISP values make any sense if we're pretending LiquidFuel is hydrazine or something and Oxidizer is N204? Could the ISP for LF/OX be lowered further? Wikipedia tells me that the proton rocket hypergolic first stage gets 285s, and the upper hypergolic stages get ~325s. 370 is pretty high. I haven't actually tried this mod yet so I don't know how the numbers add up for all the different engines and fuels...

[Legwan]

First I'd like to point at rocket equation - with such a light fuel tank can easily lead to a situation when payload (i.e. capsule) can make up a significant part of the overall mass (while it should be in the range of 30-35%).

Quick numbers: mk1-2 pod 4.12t + orange tank 3.81t (!!! out of which 2.85t is fuel) + nerva 2.33t which giver mass ratio of mere 1.38 (this goes into ln() in rocket equation). Note that pod here is 67% of total mass.

Think about it - to make it work a single mk1-2 pod needs at least 2 giant orange tanks and a gray one to be somewhere close to mass distribution as in Apollo era! Not even talking about additional parts there (like decoupler)

And this way we arrive to the main topic of fuel tanks. Unfortunately the mod is correct with their structural mass and we shouldn't attempt to

renounce reality here.

To support my words lets use space shuttle External Tank, which has been used as baseline for ksp orange tank.

Since we're talking about NERVA so only H2 will be taken account, so using data from NASA webpage you can estimate than mass of ET filled with H2 (no LOX) would be 291k pounds while dry mass is 78k pounds -> mass ratio of 3.73

For comparison KSP orange tank: full H2: 3.81t, dry 0.96 -> mass ratio of 3.96

Since ET with H2 would probably would lose some structural weight those two ratios would be even closer.

But that's not the whole terrible story! Another engineers nightmare is volume-to-surface-area (V/A) ratio for given body. In great simplification structural mass is proportional to overall surface area of container, while mass of stored hydrogen is proportional to volume. In case of cylinder this ratio increases as diameter increases:

ET: diameter 8.4m, height 47m -> V/A = 6.4

Ariane5 Cryo stage: diameter 5.4m, heigh 30.5m -> V/A = 2.9

KSP orange: diameter 2.5m, height scaled ~14m -> V/A=0.7 (!!!!)

That means the same weight of H2 stored in orange tank will take more than 9x more structural mass as compared to ET. And as we know thanks to rocket equation structural mass is bad. No space for kerbals anymore.

Thankfully this aspect is not incorporated in this mod (orange tank has volume 6400 instead of ~68 as it should have) - number has been choosed such that fuel tanks still behave as their real life counterpart in terms od mass ratio (total/dry).

So now to the topic of nerva vs. others. Wanted to write another lenghty pharagraph here but I've found it nicely described in wikipedia under this link

To summarize:

NERVA powered stages will perform much worse if we only swap the engine, despite higher Isp. In KSP even after adding volume to mach the original mass of the stage the performance (dV) will be lower because structural mass is assumed additive (you just put additional tanks on). However in real life you'll build a bigger tank, which thanks to V/A ratio will have significantly lower structural mass. This in turn will profit in rocket equation to such extent that NERVA will be more efficient than chemical stage.

Honestly doesn't have much idea how to fix it in KSP... One thing for sure - proper fix won't make NERVA super efficient, rather in-par with other engines.

Edited August 3, 2013 by Legwan

[pina_coladas]

^^^I stand corrected about the fuel tanks!

[The Lone Wolfling]

Yes.. the tank mass ratio can in fact have something to do. But the huge difference that we saw in performance it need to have another cause.

Does it? Because the numbers I get in-game match with the numbers I get by manually calculating the mass ratio.

Something tells me that KSP is converting only the half of the fuel into thrust when it use only H2

We have the normal fuel ratio of 0.9 and 1.1 for oxidexer that the game usual use. So... Is possible that when the mod tells to ksp that it has only H2 to burm, then it use 0,9 of H2 and 1.1 extra of H2 to get the same thrust? So it will burn 2 times the H2 needed?

Nope. The engine calculates a mass flow rate and backtracks that to propellant amounts based on the ratio of the fuels. It's burning the "right" amount, AFAIK.

Well, do the stock ISP values make any sense if we're pretending LiquidFuel is hydrazine or something and Oxidizer is N204?

They are high I believe- see here. ~286s at sea level.

Could the ISP for LF/OX be lowered further? Wikipedia tells me that the proton rocket hypergolic first stage gets 285s, and the upper hypergolic stages get ~325s. 370 is pretty high. I haven't actually tried this mod yet so I don't know how the numbers add up for all the different engines and fuels...

Yes it could, but then you're making it fall out of line with stock.

First I'd like to point at rocket equation - with such a light fuel tank can easily lead to a situation when payload (i.e. capsule) can make up a significant part of the overall mass (while it should be in the range of 30-35%).

Quick numbers: mk1-2 pod 4.12t + orange tank 3.81t (!!! out of which 2.85t is fuel) + nerva 2.33t which giver mass ratio of mere 1.38 (this goes into ln() in rocket equation). Note that pod here is 67% of total mass.

Think about it - to make it work a single mk1-2 pod needs at least 2 giant orange tanks and a gray one to be somewhere close to mass distribution as in Apollo era! Not even talking about additional parts there (like decoupler)

Again, even assuming an infinite mass of fuel tanks (or equivalently payload + engine weight = 0) it's still not worth it currently.

And this way we arrive to the main topic of fuel tanks. Unfortunately the mod is correct with their structural mass and we shouldn't attempt to

renounce reality here.

To support my words lets use space shuttle External Tank, which has been used as baseline for ksp orange tank.

Since we're talking about NERVA so only H2 will be taken account, so using data from NASA webpage you can estimate than mass of ET filled with H2 (no LOX) would be 291k pounds while dry mass is 78k pounds -> mass ratio of 3.73

For comparison KSP orange tank: full H2: 3.81t, dry 0.96 -> mass ratio of 3.96

Since ET with H2 would probably would lose some structural weight those two ratios would be even closer.

Huh. I've seen the H2 mass ratio quoted as higher for tug designs, but those don't need to hold up under 1G. Although that being said I wouldn't apply IRL physics to KSP - the dense planets / atmospheres throws everything off.

So now to the topic of nerva vs. others. Wanted to write another lenghty pharagraph here but I've found it nicely described in wikipedia under this link

To summarize:

NERVA powered stages will perform much worse if we only swap the engine, despite higher Isp. In KSP even after adding volume to mach the original mass of the stage the performance (dV) will be lower because structural mass is assumed additive (you just put additional tanks on). However in real life you'll build a bigger tank, which thanks to V/A ratio will have significantly lower structural mass. This in turn will profit in rocket equation to such extent that NERVA will be more efficient than chemical stage.

Honestly doesn't have much idea how to fix it in KSP... One thing for sure - proper fix won't make NERVA super efficient, rather in-par with other engines.

Square-cube law, in other words. Perhaps what we need is to scale tank internal volumes to assume a thickness of tank wall proportional to the radius?

Edited August 3, 2013 by The Lone Wolfling

[pina_coladas]

Yes it could, but then you're making it fall out of line with stock.

Might as well mess with the stock values if it helps the whole mod make more sense IMO.

And it has always bugged me a bit that larger tanks in KSP don't have better mass fractions due to increased volume to area ratios. It would be cool to do something about it...

Edited August 3, 2013 by pina_coladas

[The Lone Wolfling]

Here is my first attempt at a MFT config pack that makes other configs besides LF/OX actually useful. It only has the squad parts, and it still needs a bunch of tweaking (in particular I should scale the atmospheric Isps by the ratio of the old vacuum Isp to the new vacuum Isp instead of using the scaling of LF/OX), but I might as well get other eyes on it.

[pina_coladas]

I can't try it out just yet but I'll give it a look at some point. Thanks for putting it together.

[AngelLestat]

I was doing more test, and it seems that you are right Lone wolfling, the dry tank mass is the problem here.

It was hard to see it becoz I was using big tanks from the nova punch pack and when you use big tanks the effect increase, so I had a craft with 44 tons full weight and without all the H2 was 20 tons.. The half of the weight was dry mass.

[s20dan]

Here is my first attempt at a MFT config pack that makes other configs besides LF/OX actually useful. It only has the squad parts, and it still needs a bunch of tweaking (in particular I should scale the atmospheric Isps by the ratio of the old vacuum Isp to the new vacuum Isp instead of using the scaling of LF/OX), but I might as well get other eyes on it.

I tried these settings. They just make things way too easy.

I strongly get the impression there is an issue with how the game calculates fuel mass flow.

LOOK:

2 identical craft, same fuel MASS, same payload mass, same tank sizes. (BOth after tinkering weigh 51 tons, LF/OX has some fuel drained out.)

EXCEPT: one uses LF/OX the other uses LF/LOX... LF/LOX has a higher TWR (sometimes) and a higher ISP so WHY THE **** DOES IT HAVE A LOWER DELTA-V?? (The anger is supposed to emphasise my confusion )

Both craft have same dry weight, same fuel mass ect.. what the hell is going on? Just do this test yourself.

Edited August 4, 2013 by s20dan

[xZise]

[…]Wrong. IRL mass ratios are much higher.[…]

Okay I was talking about the calculation KSP is doing. I also looked the density of liquid hydrogen up and get 72.01 kg/m³. Now the ResourcesFuel.cfg states that the unit is kg/6.25 m³ which I highly doubt that's correct, as the given values are way off. For example liquid fuel has a density of 0.005 u but according the density of RP-1 is about 1 g/ml (= 1 g/cm³ = 1 kg/dm³ = 1000 kg/m³). It looks more like Mg/6.25 dm³ (t/6.25 l), which would get me a density of 0.000011 u (≃ 0.00007201 Mg/dm³). Which is way below the density given in the configuration file and actually make the complete situation worse. Or maybe I did a conversion error because LOX has a density of 1146 kg/m³. With LH2/LOX density ratio of 0.062) I get 0.062·0.007=0.000434 (0.007 u is the density of LOX in the configuration file) which matches almost the given density of LH2. But then again RP-1 is either to light or LOX to heavy (and then LH2 too), as 1146 kg/m³/1 g/ml·0.005=0.00573 (0.005 u is the density of RP-1/LF in the configuration file) this would be density of LOX.

[…]Wrong again. We know how much fuel a NERVA would need as we've conducted tests. Not to mention that it's pretty easy to calculate approximate exhaust velocities of a purely thermal rocket, all things considered.[…]

My wording on this was poor, but saying my rocket looks unrealistic because it's so huge doesn't really work, as we never saw a rocket with one NERVA engine. They only conducted ground based tests so maybe a rocket with a NERVA engine simply looks this way.

[…]Something needs to be done about the LV-N. Currently with the real fuels it is never worth it to use a LV-N.

Yeah I was interested too what the problem could be. And it looks like you make some progress in making LOX and LH2 useful.

Fabian

[xZise]

I tried these settings. They just make things way too easy.

I strongly get the impression there is an issue with how the game calculates fuel mass flow.

LOOK:

2 identical craft, same fuel MASS, same payload mass, same tank sizes. (BOth after tinkering weigh 51 tons, LF/OX has some fuel drained out.)

EXCEPT: one uses LF/OX the other uses LF/LOX... LF/LOX has a higher TWR (sometimes) and a higher ISP so WHY THE **** DOES IT HAVE A LOWER DELTA-V?? (The anger is supposed to emphasise my confusion )

Both craft have same dry weight, same fuel mass ect.. what the hell is going on? Just do this test yourself.

I'll test that later today, but who calculates you the ÃŽâ€V? As I already said I'm pretty confident that the calculations KSP is doing are correct.

Fabian

[The Lone Wolfling]

I tried these settings. They just make things way too easy.

Agreed, unfortunately. The issue is that as the size increased the advantage swings towards the fuel with the higher mass ratio asymptotically, which means that in order to get the very large mass case right it means that at small sizes the lighter fuels are much better than the denser fuels.

It looks like I'm going to next try reducing dry tank weight instead and seeing if that helps.

I strongly get the impression there is an issue with how the game calculates fuel mass flow.

LOOK:

2 identical craft, same fuel MASS, same payload mass, same tank sizes. (BOth after tinkering weigh 51 tons, LF/OX has some fuel drained out.)

EXCEPT: one uses LF/OX the other uses LF/LOX... LF/LOX has a higher TWR (sometimes) and a higher ISP so WHY THE **** DOES IT HAVE A LOWER DELTA-V?? (The anger is supposed to emphasise my confusion )

Both craft have same dry weight, same fuel mass ect.. what the hell is going on? Just do this test yourself.

Moment.

I cannot reproduce. What are you using to calculate delta-v? Are you sure that you removed the same amount of LF and OX from the tank? Are you sure that the tank fuel flow is correct? Can you show pictures / a craft file?

[s20dan]

I'll test that later today, but who calculates you the ÃŽâ€V?

I cannot reproduce. What are you using to calculate delta-v? Are you sure that you removed the same amount of LF and OX from the tank? Are you sure that the tank fuel flow is correct? Can you show pictures / a craft file?

Good point. I'm just using Mechjeb and the Kerbal Engineering System, there could very well be a problem there.

The test was something like this:

The LF/OX actually has less fuel mass than the LF/LOX rocket as it's carrying some dead weight (it was easier to get the desired mass that way) and it still comes out on top.

(Oh I realise that's a mod tank, the result is the same with the Jumbo orange tank).

[xZise]

I didn't check it with my own calculations, but I guess it's because of the additional dry mass. You could test if you add enough fuel to the LF/OX combination which isn't LF/OX so that the dry mass (the mass of the tank fully fuelled except for LF and OX) is about the same dry mass as the LF/LOX tank.

Or another possibility is, that KER/MJ don't really bother what type of fuel you use. So if you fill the tank only with LF if KER and MJ still calculate the ÃŽâ€v as if you could use all LF (what you can obviously as oxidizer is missing).

At the moment I can't test it on my own so I can only suggest.

Fabian