[WIP] Nert's Dev Thread - Current: various updates

[Nertea]

It's right above, in the same post. 

[Streetwind]

Just now, Nertea said:

It's right above, in the same post. 

...The heck. So it is.

Okay, that seals it. I am going to bed right now. x___x

[billkerbinsky]

1 hour ago, Psycho_zs said:

It would be strange to drop hydrogen. It has its problems, but usually whey you hear "NTR", you think "hydrogen!".

And I have a very warm memories of a swift argon-fueled VASIMR-driven research vessel strapped to a monstrous interplanetary hydrogen 'backpack'. It zipped to Jool in almost a straight line, spending less than a third of fuel. It was a pain escaping Kerbin though. But Persistent Thrust mod may fix that.

Likewise.   I built a fueling station out of a bunch of those ridiculous 10m spherical LH2 tanks...   

[Fraz86]

4 hours ago, Nertea said:

Xe tanks normalized to a dry mass ratio of 1.28 (same as large stock tank)

I just wanted to point out that the PB-X750 (the large stock tank) actually has the worst dry mass ratio of the stock Xe tanks. The dry mass ratios of the PB-X50R and PB-X150 are 1.333 and 1.4, respectively.

[Captain Sierra]

Will grab and test as soon as I see a DL. Also Nertea, I am working on a quick and dirty MM patch to make your radiators work like stock so that I can still operate NFE without the eyesore stock radiator panels. It may make a viable stopgap solution until a proper handling of HC can be completed.

[Streetwind]

7 hours ago, Fraz86 said:

I just wanted to point out that the PB-X750 (the large stock tank) actually has the worst dry mass ratio of the stock Xe tanks. The dry mass ratios of the PB-X50R and PB-X150 are 1.333 and 1.4, respectively.

Erm... where did you see that?
The PB-X50R has a dry mass of 31.43kg and contains 40.00kg of fuel. Thus 40 / 31.43 = 1,2727
The PB-X150 has a dry mass of 55.00kg and contains 70.00kg of fuel. Thus 70 / 55 = 1.2727
The PB-X450 has a dry mass of 412.50kg and contains 525.00kg of fuel. Thus 525 / 412.5 = 1,2727

Looks pretty identical to me. The trap is in the way KSP's UI rounds masses down to the nearest 10kg, showing just 0.07t wet mass for the X50R and 0.12t wet mass for the X150. But if you check the config files you can see the real dry mass.

Nertea probably made the same mistake and took 410kg dry mass for the X450, resulting in a mass ratio of 1.280. But eh, close enough

 

Edited December 9, 2015 by Streetwind

[Nertea]

Let me know how this goes...

https://www.dropbox.com/s/2dsjpuaxs4qqno6/NearFuturePropulsion0_6_T1.zip?dl=0

 

 

[Streetwind]

6 minutes ago, Nertea said:

Let me know how this goes...

https://www.dropbox.com/s/2dsjpuaxs4qqno6/NearFuturePropulsion0_6_T1.zip?dl=0

Will do once I get home from work... in 9 hours...

[Captain Sierra]

7 hours ago, Nertea said:

Let me know how this goes...

https://www.dropbox.com/s/2dsjpuaxs4qqno6/NearFuturePropulsion0_6_T1.zip?dl=0

Oh awesome you caught me on my day off. Time to field test! Will toss you a PM later today or tomorrow with bug and balance reports.

[Streetwind]

Thoughts and notes on the test version:

- Dawn patch is broken. Reason: Dawn uses ModuleEnginesFX by default now, patch is still trying to target and convert ModuleEngines.
- Dawn now has a stock exhaust effect. Replacing it with Near Future's still works, but is it necessary? Your call.
- VASIMRs in xenon mode have wrong Ec/s numbers. Filing this under "as expected", considering the density is different from hydrogen, which means the Ec propellant ratio is now off.
- Lithium tanks fragile? Yep. Flew an aggressive ascent, tank was by far the first part to go. Flew a normal ascent, tank survived. Means you can get them into space fine, but aerobraking or reentry with these will probably be very difficult.
- If lithium is supposed to be the extremely cheap fuel with extremely expensive tanks, it might make more sense to have a different fuel with the fragile tankage, since this setup encourages making efforts to refuel an existing spacecraft with a reusable vehicle that must reenter safely to be recovered.
- MPDTs work fine with lithium as far as I can see.
- Mass ratios: A penalty of ~0.03 for argon tanks will not do anything noticable. For reference, the performance difference between 1.4:1 and 2.0:1 ends up being about 4%... you can imagine how big the difference between 1.25:1 and 1.28:1 will be Even if you only want a very small penalty, this one is so small as to be negligible. Also, I'm not sure I'm a fan of the concept of penalizing argon tank mass ratio to make up for the fuel's cost; I'd rather adjust the engines (or rather, buff the xenon ones a bit). With the KSPedia to inform the player, IMHO that will be the more intuitive option for handling things. Players will look at it like xenon is the higher priced performance fuel at the gas station that makes your car run better.
- Mass ratios part 2: For the same reason as given above, lithium tanks won't feel markedly different unless the difference in mass ratio is larger. Something like 2.5:1 if xenon and argon stay with stock numbers.
- Mass ratios part 3: is staying with stock numbers for xenon and argon something you fancy, or was that just the easiest option for the test version?

Cost analysis:
Xenon: 40 funds/kg fuel; xenon tanks: 2.0-2.5 funds/kg fuel.
Argon: 5.886 funds/kg fuel; argon tanks: 1.4-2.8 funds/kg fuel.
Lithium: 0.506 funds/kg fuel, lithium tanks: 4.21 funds/kg fuel.

EDIT: I really, really need to stop posting late at night and making myself look dumb *quietly hides embarrassing paragraph in a trash bin far, far away*

 

...okay, I think I'm done for today.

 

Edited December 9, 2015 by Streetwind

[Nertea]

That's an easy fix at least. I will not change the exhaust :). The new Dawn one is closer to what I envision for v2 of most of my effects. 

[Psycho_zs]

Shoulnd't VASIMRs have less ISP with xenon than argon?

[billkerbinsky]

50 minutes ago, Psycho_zs said:

Shoulnd't VASIMRs have less ISP with xenon than argon?

That's certainly the case for thermal engines -- for a given temperature, lighter-molecular-weight propellants move faster.   (Isp is directly proportional to exhaust velocity).

If Xe has an advantage I'd think it would be tank volume per mass of propellant.
 

[Captain Sierra]

1 hour ago, Psycho_zs said:

Shoulnd't VASIMRs have less ISP with xenon than argon?

 

26 minutes ago, billkerbinsky said:

That's certainly the case for thermal engines -- for a given temperature, lighter-molecular-weight propellants move faster.   (Isp is directly proportional to exhaust velocity).

If Xe has an advantage I'd think it would be tank volume per mass of propellant.
 

 

Provided the delta V still goes up for the same volume of xenon as compared to argon, than balance-wise there's nothing wrong with this, but I would like to make the point that as with anything, it should be balanced for gameplay first, realism second.

[Nertea]

12 hours ago, Streetwind said:

Thoughts and notes on the test version:

- Dawn patch is broken. Reason: Dawn uses ModuleEnginesFX by default now, patch is still trying to target and convert ModuleEngines.
- Dawn now has a stock exhaust effect. Replacing it with Near Future's still works, but is it necessary? Your call.
- VASIMRs in xenon mode have wrong Ec/s numbers. Filing this under "as expected", considering the density is different from hydrogen, which means the Ec propellant ratio is now off.
- Lithium tanks fragile? Yep. Flew an aggressive ascent, tank was by far the first part to go. Flew a normal ascent, tank survived. Means you can get them into space fine, but aerobraking or reentry with these will probably be very difficult.
- If lithium is supposed to be the extremely cheap fuel with extremely expensive tanks, it might make more sense to have a different fuel with the fragile tankage, since this setup encourages making efforts to refuel an existing spacecraft with a reusable vehicle that must reenter safely to be recovered.
- MPDTs work fine with lithium as far as I can see.
- Mass ratios: A penalty of ~0.03 for argon tanks will not do anything noticable. For reference, the performance difference between 1.4:1 and 2.0:1 ends up being about 4%... you can imagine how big the difference between 1.25:1 and 1.28:1 will be Even if you only want a very small penalty, this one is so small as to be negligible. Also, I'm not sure I'm a fan of the concept of penalizing argon tank mass ratio to make up for the fuel's cost; I'd rather adjust the engines (or rather, buff the xenon ones a bit). With the KSPedia to inform the player, IMHO that will be the more intuitive option for handling things. Players will look at it like xenon is the higher priced performance fuel at the gas station that makes your car run better.
- Mass ratios part 2: For the same reason as given above, lithium tanks won't feel markedly different unless the difference in mass ratio is larger. Something like 2.5:1 if xenon and argon stay with stock numbers.
- Mass ratios part 3: is staying with stock numbers for xenon and argon something you fancy, or was that just the easiest option for the test version?

Cost analysis:
Xenon: 40 funds/kg fuel; xenon tanks: 2.0-2.5 funds/kg fuel.
Argon: 5.886 funds/kg fuel; argon tanks: 1.4-2.8 funds/kg fuel.
Lithium: 0.506 funds/kg fuel, lithium tanks: 4.21 funds/kg fuel.

EDIT: I really, really need to stop posting late at night and making myself look dumb *quietly hides embarrassing paragraph in a trash bin far, far away*

 

...okay, I think I'm done for today.

 

Hehe, I saw your embarrassing paragraph and was quite confused. Okay, let's see...

Yeah I might have been a tad to aggressive on the heat tolerance front with the Li tanks. The cost might need to go up a bit, as far as I know though KSPI-E is the Li curator in CRP so I have to get that cleared. It shouldn't be that cheap compared to the others.

I do prefer to stay with stock numbers for Xe mass ratio. I am pleased that we managed to only change one stock thing in the whole set of mods and I don't want to change more parts if it can be avoided. 

6 hours ago, Psycho_zs said:

Shoulnd't VASIMRs have less ISP with xenon than argon?

Probably - as you can probably note, I changed absolutely nothing about those cfgs except for the propellant name. What they should really have is more system efficiency as well, so I was thinking that if this is the way to be going, Xe mode would have less heat generation (75% efficient instead of 55%) and a bit lower specific impulse. I would appreciate Streetwind's advice on how this might look numerically.

 

[Streetwind]

Re: Isp with xenon.

No, the high Isp is correct. As billkerbinsky noted, it would apply to thermal engines, but the VASIMR is not a thermal engine. It's an electric engine.

Electric engines work (broadly speaking) by expending large amounts of electric power into their propellant, giving the atoms a charge and turning them into ions. These ions are susceptible to electromagnetic forces, and thus, the engine can produce such a field to push against the ionized propellant, shooting it away from itself to generate thrust.

So why use xenon, of all things - an absurdly expensive element? This is not by mistake, or because someone trolled their fellow engineering colleagues. No, xenon is simply the ideal electric engine propellant. It has a huge amount of electron shells, making it very easy to ionize - the xenon atom more or less doesn't resist this process, in contrast to smaller atoms which hold their electrons much more tightly. That means the engine has to put hardly any energy into the ionizing phase, and can put nearly all of it into the electromagnetic pushing phase. A propellant like argon is way cheaper, but you have to fight the argon atom to put it into the same ionized state. Thus you have to expend more energy on that process, and have less left for pushing; your engine performance suffers.

As a result, xenon is used widely, while other propellants see only niche use. In a spacecraft, the propellant cost is usually negligible, even at the price point of xenon - especially if you need only a few dozen kilograms of it, since xenon generates such high specific impulses for the amount of energy invested.

It's only with engines like the VASIMR, which is expected to run at up to a hundred times the input power of contemporary electric engines and to be refueled multiple times over its lifetime, that it actually makes sense to pay attention to the price of the propellant. That's why the IRL VASIMR runs on argon. It could, however, run on other propellants just as well, including xenon. And the engine performance will depend on how well each propellant is suited to electric engines.

(I'll note here that I'm by no means an electric propulsion engineer, and so I can only give a layman's explanation which may not be 100% technically correct. But xenon is indeed preferred in many engine types for its near-ideal propellant behavior - you can look this up all over the place, and I invite you to do so because it's a really interesting topic. At least to me, that is. )

 

Edited December 10, 2015 by Streetwind

[Nertea]

Streetwind, that may not be correct. Check this paper (p11), where it indicates that increasing the atomic mass decreases specific impulse http://www.umich.edu/~peplweb/pdf/AIAA-2012-3930.pdf (am i reading this correctly?). However lower ionization energy increases system power to jet efficiency, which would give more thrust and less wasted power. Xenon isn't represented but the relationship between Ar and Kr should be similar to that of Kr and Xe.

[Psycho_zs]

A quick search yielded this, it also states that lighter propellants are preferable, at least for long range missions.

So, xenon would be more efficient in electrical department - allows for lower power consumption and/or lower heat generation, argon would give higher ISP. I wonder how this would balance out. If you've got some spare power, could you use it to propel ions faster?

P.S. other VASIMR FAQ answers at ADAstra site have some useful info.

Edited December 10, 2015 by Psycho_zs

[Streetwind]

Well, the FAQ does point out that the VASIMR does its ionization a little differently than other electric engines... perhaps in a way that benefits lighter gases? I don't know. My post was aimed at electric engines in general, not just the VASIMR; it's possible that the differences in engine technology makes for example gridded ion thrusters react differently to xenon and argon than the VASIMR does.

@Nertea, interesting paper. I don't pretend to understand everything that's in it, but the graph on page 11 is nice. Too bad that the lines aren't fully etrapolated, I would really like to know if the lines for lighter gases end up crossing those of heavier ones eventually.

I can try and come up with some new numbers for the xenon mode. Regardless of how it's going to turn out, though, it'll have to end up with a higher overall engine rating than the argon mode, to account for the cost difference. Not much - I'm thinking 5%-10% here - but there's gotta be some reward for paying the extra buck...

How do you feel about less total power input for the xenon mode? Would let people get away with smaller power production solutions in theory, improving TWR. In practical application, not sure such a thing will work out, because there's only a few reactors in a few sizes available in NF Electrical, and you cannot just downsize easily. So yeah, it's an option, but I could also balance just by thrust/Isp differences as it is set up now.

One reason to keep xenon as the high Isp mode on the VASIMR, by the way: all other xenon engines are the high Isp options of their respective tiers.

Edited December 10, 2015 by Streetwind

[Nertea]

Let's try what I've done here - significantly less heat production for Xe mode.

Here's two things:

New NFP build 

Should fix the Dawn patch, tweak the mass ratio of Ar tanks, puts the Xe VASIMRs to 1.4x the thrust and 1/1.14x the Isp with 36% heat savings 

NFE build

This is *not* perfect. There are a number of minor changes here. The first is that the power slider is now a "reaction rate" slider which causes the core to heat up. Power generation is scaled by the core heat - at full temperature the power generated = the power created. Radiator capacity required is now equal to the waste heat, which is shown in the Core Temperature field. You match the required number to the capacity of radiators. This is in fact pretty easy to do, which is a huge improvement. Thus the system is both more complicated and less complicated than it used to be. In addition, due to their naturally poor heat transfer versus their size, the stock radiators are okay at doing this, but aren't the most efficient at it. This leaves room for high temperature radiators, so HC has a nice easy niche to fill soon. Fun!

The biggest problem here is that you'll see that the power setting is kinda annoying to use. Varying the reactor power doesn't do as much as I like in terms of managing the core temperature - you have to have it set to like 1-5% in order to cool it down for example. I don't have a full handle on how the core heat diffusion works. This isn't too horrible - the worst that could happen is that I remove the slider and just have a reactor on/off button. Could be fine. A second option is to have the reactor power slider vary the core heat target between ambient and nominal temperatures, thus allowing power adjustment (radiators just keep core heat below the heat target). Please consider both these options :).

Another minor problem is that the right click menu is polluted, messy and mislabeled. That I can fix without too much trouble.

Edited December 10, 2015 by Nertea

[Captain Sierra]

So I took some time to do some reading because VASIMRs are without a doubt the most complicated propulsion mechanism ever to be modeled in KSP and Nertea has done one of the more realistic jobs of it.  Here's what I think after brushing up on how these things work:

The current thrust and heat improvements are good. Those are sufficiently realistic. The ISP changes are not realistic but I dont believe them to be a problem. All other xenon engines have higher ISP than comparable argon engines. Having the VASIMR be the exception to that rule may be unintuitive to the player (and this mod already can be without adding to it). I think its worth it to take a pass on being realistic there and instead of taking like a 10% drop to ISP you leave it as is, very slightly buffed.

 

I'm downloading both now with the intent of testing out. Will report back with an edit after experimenting with the test builds. Also my progress has hit a stumbling block on a hack patch to make Nertea's radiators use stock function (I can just barely tolerate them functionally, but cant stand them visually, so this makes it bearable).

[Psycho_zs]

Getting busy testing...

New eactor models look fantastic!

MX4 reactor: both nodes are offset. These stats seem to fit nicely:

	node_stack_top = 0.0, 1.585, 0, 0.0, 1.0, 0.0, 1
	node_stack_bottom = 0.0, -0.705, 0, 0.0, -1.0, 0.0, 1

 

Edited December 10, 2015 by Psycho_zs

[billkerbinsky]

1 hour ago, Captain Sierra said:

I think its worth it to take a pass on being realistic there and instead of taking like a 10% drop to ISP you leave it as is, very slightly buffed.

Getting the sign or slope or relative difference wrong breaks immersion for me in a way that getting the absolute magnitudes wrong doesn't.   All else being equal, I'd rather see the relative I_sp of heavier molecular weight propellents in the same engine be lower, with Xe winning out for other reasons when it makes sense.

[Streetwind]

4 hours ago, Nertea said:

NFE build

This is *not* perfect. There are a number of minor changes here. The first is that the power slider is now a "reaction rate" slider which causes the core to heat up. Power generation is scaled by the core heat - at full temperature the power generated = the power created. Radiator capacity required is now equal to the waste heat, which is shown in the Core Temperature field. You match the required number to the capacity of radiators. This is in fact pretty easy to do, which is a huge improvement. Thus the system is both more complicated and less complicated than it used to be. In addition, due to their naturally poor heat transfer versus their size, the stock radiators are okay at doing this, but aren't the most efficient at it. This leaves room for high temperature radiators, so HC has a nice easy niche to fill soon. Fun!

The biggest problem here is that you'll see that the power setting is kinda annoying to use. Varying the reactor power doesn't do as much as I like in terms of managing the core temperature - you have to have it set to like 1-5% in order to cool it down for example. I don't have a full handle on how the core heat diffusion works. This isn't too horrible - the worst that could happen is that I remove the slider and just have a reactor on/off button. Could be fine. A second option is to have the reactor power slider vary the core heat target between ambient and nominal temperatures, thus allowing power adjustment (radiators just keep core heat below the heat target). Please consider both these options :).

Another minor problem is that the right click menu is polluted, messy and mislabeled. That I can fix without too much trouble.

Okay, I am super confused about this new reactor thing. On the risk of making a late night fool out of myself for the third day in a row - there must be something wrong here...

The first thing I did after downloading was to grab a small raft with a reactor and a VASIMR and throw it into orbit. The M-EXP makes an amusing, if easily damaged, rocket nosecone Here is a screenshot. Since the M-EXP called for 150-200 kW of radiator power in the VAB, I packed four small thermal control systems (50 kW core transfer each).

Which means you should immediately notice something odd, which is the very thing that is confusing me. Currently, I only have one radiator extended. And yet, the M-EXP is perfectly stable at 650K/650K, 100% core health, and generating its full 50 Ec/s. This is not a moment snapshot, this is steady state.

I've done some experimentation, reset a few times, and ultimately came to the following conclusions:
- RadUsage and RadCoolAmt show the amount of core transfer available. Not sure of the difference between the two, but they always show identical numbers.
- RadCoolTot appears to scale directly with the reactor control slider. In the screenshot, at 33%, it is at 47.7, which is less than 50, thus the reactor remains stable. If I would turn it any higher, RadCoolTot would exceed 50 and thus the reactor would start gaining core temperature. Interestingly, RadCoolTot never shows more than RadCoolAmt shows, which probably means that this number describes the total amount of core heat being transferred out of the part at any given time (which obviously cannot exceed the amount of transfer available). Though if no heat is transferred at all, this value doesn't become 0, but rather equal to RadCoolAmt? Super confusing. It also appears that this amount of heat transfer scales itself to what the reactor currently produces. I can for example turn the reactor control slider all the way down to 5%. That is the minimum setting. RadCoolTot shows 5.7 at this point.
- The strange thing here is this: even at that setting, the reactor remains stable at maximum heat, and generates maximum power output. It does not cool down, because the amount of heat transfer throttled itself automatically to stop the reactor from cooling down.

This reliably lets me go all exploity and do this. Yep, this is a MX-L reactor, outputting 3,000 Ec/s steady state... while being cooled by three small radiators. Total heat throughput: 144.7 kW out of 150 kW capacity. If I could turn it down all the way to 1%, it would even work with one small radiator. (Amusingly, fuel usage seems to scale off of temperature, not the reactor control slider, meaning that fuel usage is actually the proper 100% burn rate, despite the reactor being set to 5%.)

This leads me to believe that power created depends on nothing but current core temperature. Any setup that will hold that temperature steady state gains full output. Unfortunately, this "any setup" includes those which do not transfer anywhere near the total amount of heat throughput the reactor is able to create. It works because you can simply throttle down. This is why, Nertea, you are finding that the slider doesn't really affect what the reactor is doing unless you move it down to essentially zero.

Suggestion on how to fix this:
- In the real world, a heat engine's energy throughput is defined by the amount of heat it moves from A to B. Therefore the power output of a generator connected to a heat engine is also defined by the amount of heat the heat engine moves from A to B. Where A is the heat source and B is the heat sink. In other words: the reactor's power output should scale linearly with the amount of heat transferred away from it, into the radiators. That means you always only ever get the Ec output for which you have radiator power onboard.
- At the same time, fuel usage must scale with the reactor control slider, if it is really meant to signify the reaction rate. That you burn more fuel with a higher reaction rate only makes sense, yes?
- And what does core heat do, then? It could define the rate at which that heat transfer to the radiators happen. Right now, a reactor below nominal temp transfers nothing whatsoever to the radiators, but a reactor at nominal temp transfers everything it currently produces (and one above nominal will also transfer excess to cool itself down). This sudden 0 to 100 isn't very satisfying. It would be nicer if the radiators would start transferring even below nominal, just not at the full rate. For example, a linear increase over the temperature spread from 400 K (arbitrary minimum heat to generate power) to whatever the nominal temp is. For a M-EXP, 400 K to 650 K (a range of 250), where at 400K it would transfer 0% of the current output to the radiators, at 525 K it would transfer 50%, and at 650 K it would send 100% of the heat output into the radiators. This would result in the reactor temperature rising quickly at the start, but growing more slowly the closer it comes to nominal. The amount of power produced would do the same.

DISCLAIMER: I don't know if you can actually mod these systems like that. The above is just how I would do it if I could.

 

And there's something else that I would like to mention:

When coming out of a scene change to a deactivated reactor, all of the debug numbers only show questionmarks. If you turn on the reactor, it is supposed to detect that there are radiators available - at least if you have any that are currently working - and fill in those numbers with the relevant info. Even if no radiators are available at the moment of reactor start, the reactor should notice when radiators are toggled on afterwards. This works just fine if you for example switch to an active vessel from the tracking station.

However, when the reactor comes fresh out of the VAB, and was never on before, this does not work reliably. The very first time you activate the reactor, it starts and begins to heat up, but it does not resolve the questionmarks. Sometimes waiting helps, but that is dangerous, because the reactor may reach full meltdown before anything happens. Sometimes turning it off and back on again helps (thank you tech support!), but not always. Sometimes toggling a radiator helps, but not always.

As soon as the reactor properly resolved the questionmarks once, you can turn if off and change scenes as much as you like; whenever you come back, it will reliably identify its cooling situation. Just not that one first time, when the vessel is brand new.

 

Edited December 10, 2015 by Streetwind

[Psycho_zs]

Flying a craft with MX-3 reactor and WV-200 engine..

A 800K core heat reactor produces 363,5 Ec/s.

I see no change in heat flow that comes through radiators on reactor setting from 6% to 65% (using 2xMedium radiators). Above 65% the core quickly overheats.

Either I'm missing something here, or something is wrong. If between 6% and 65% "reaction rate" core temp stays at 800K and radiators pump the same amount of heat, then two questions remain. What this setting do in that range, and where does excess heat go? I see RadCool, RadXfer, RadExcess values jump instantly on reactor rate changes, but no parts heat up or cool down.