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

[Nertea]

New NFP: Xe VASIMR effects, Li MPDT effects, fixed Xe mode VASIMR labeling, added 2 1.25m Li tanks. 

New NFE: Better throttling of electricity at lower powers. Actually works in timewarp now. Better stability/equilibrium points at low powers. Core damage is back (but may change). Still some bugs, but getting there (note that my flagship test is the 2.5m reactor, I don't really use the others for quick tests). 

13 minutes ago, Captain Sierra said:

Some notes I'd like to quickly make.

I find reactor throttle mostly unnecessary. It serves to give me a new way to screw up while not being all that useful. A simple on/off switch would be just fine (maybe a hibernate/idle mode but technically unnecessary), at least IMHO. Also I think Psycho is onto something. The simpler the function as well as the code behind it, the better for everyone. Until the KSPedia is a thing I think this mod is nontransparent enough.  (Hell, you've thoroughly confused the crap out of me about how the reactors work now and I'm a vet of these mods.)

Lithium makes MPDTs OP AF. You can get so much more dV for so much less volume now its insane. They've just become the end-all/be-all of late game propulsion. That said, it feels .... surprisingly good. I miss the LH2 out of nostalgia but the lithium version works really well, too well honestly. Since I brought up the LH2, what's your plan for those tanks? They're far too pretty to just deprecate indefinitely. I have a few concepts bouncing around but I want to hear your opinions on it, because those matter more.

They may need some changes to their stats. It's just a fuel swap right now. LH2 tanks are almost certainly going to come back as either part of a dedicated LH2 NTRs mini-pack, or as an optional addon to CryoEngines. 

[Fraz86]

The new 1.25m Li tanks have a higher maxTemp (1200K) than the 2.5m Li tanks (800K).

[Streetwind]

11 hours ago, Nertea said:

Adding proper engine light kinda has to wait for more plugin work - the multi modes and therefore multiple colours of light. You may have historically noticed that there was never any real engine emissive on those parts.

To my great shame I must admit that I never noticed - all my VASIMR tests must have happened in sunlight

But yeah, it makes sense that it might be tricky to implement if you put it like that.

 

8 hours ago, Captain Sierra said:

I find reactor throttle mostly unnecessary. It serves to give me a new way to screw up while not being all that useful. A simple on/off switch would be just fine (maybe a hibernate/idle mode but technically unnecessary), at least IMHO. Also I think Psycho is onto something. The simpler the function as well as the code behind it, the better for everyone. Until the KSPedia is a thing I think this mod is nontransparent enough.  (Hell, you've thoroughly confused the crap out of me about how the reactors work now and I'm a vet of these mods.)

The confusing parts should soon be hidden. In the final release, you'll have a reactor throttle that controls heat production, and your reactor produces Ec depending on the amount of heat production (assuming you added the amount of radiators the VAB tells you to add) and how close the reactor is to nominal temperature. I think that's a fairly simple system to get used to once you stop worrying about how it's implemented under the hood. I also firmly disagree with "simpler = better for everyone", since the overwhelming majority of users historically has asked for far higher complexity than even this, and Nertea was all "guys I can't spend that much time just on reactors!".

 

8 hours ago, Captain Sierra said:

Lithium makes MPDTs OP AF. You can get so much more dV for so much less volume now its insane. They've just become the end-all/be-all of late game propulsion. That said, it feels .... surprisingly good. I miss the LH2 out of nostalgia but the lithium version works really well, too well honestly.

How can you say that, when in fact the lithium fuel has been an all-around nerf at the moment? The vessels are more compact, yes, but the mass ratio of the tanks has dropped from 3:1 to 1.5:1. This means you're taking twice the tankage dry mass for the same amount of fuel, which reduces your dV and your TWR. Also, tanks are more expensive than hydrogen tanks, while the fuel costs similar amounts. So overall, your lithium driven ships have less TWR, less dV, and cost more. The only "advantage" is that they look physically smaller.

But as Nertea said, we haven't even started balancing the final numbers. That's something I'm going to start poking at this weekend, when I actually have useful amounts of free time.

 

7 hours ago, Nertea said:

note that my flagship test is the 2.5m reactor, I don't really use the others for quick tests

Which one? There's two

(I'll assume the MX-1, as it has its new model already and the MX-L doesn't.)

 

Edited December 12, 2015 by Streetwind

[Psycho_zs]

MX-1 Runs really smooth! I couldn't fit all 5000kW worth of cooling power on the test ship, ended up with 3000kW, and the core temp sits still at 60% throttle.

This is good and bad. The good thing is that numbers fit perfectly and predictably. Now to the bad things:

- with fully matched cooling the core does not heat up. If I had 5000kW of radiators on the ship, I would need to disable some of them to get to nominal temp.

- Nothing can stop the runaway core.

Radiative and conductive heat loss is to the rescue! With predefined emissive (to space, the model seems OK with that) and conductive (core>part) factors it can be balanced out at the point of nominal heat, so:

[Heat production] - [radiator Cooling] - [heat Loss @ nominal temp] = 0

[Heat production] - [radiator Cooling] - [heat Loss @ suboptimal temp] = +N

[Heat production] - [radiator Cooling] - [heat Loss @ overheat temp] = -N

 

With fully matched cooling the core would slowly climb to optimal and sit there. But it would be possible to rush it up by temporarily disabling some radiators.

If cooling power is slightly missmatched, then the core would deviate from the nominal temp, but still stop at some point. Electricity penalties should be greater on the cool side.

If cooling power is greatly missmatched, hold on to yer hats!

Edited December 12, 2015 by Psycho_zs

[Streetwind]

Lithium frack yeah! 8) You can say what you want, but this is so much more badass than the hydrogen plume.

 

Meanwhile, reactors... Ec output scales with Reactor Heat Transf, which in turn scales with Core Temperature. That's perfect. But, it seems like Reactor Heat Transf so far is only a fantasy value that doesn't actually correspond to anything the reactor is doing apart from scaling with its temperature. Is that true, Nertea? Or should it work properly already and I managed to break things yet again?

 

Also, somewhere during testing, something happened to my MX-4. I have no idea why. It worked fine when I originally installed this new test release - after all I was able to generate the screenshot with the lithium plume above, which I did with a MX-4 equipped vessel. However, anytime I tried to use the MX-4 afterwards, it was broken. I went as far as deleting the entire instance and building a new one out of fresh downloads of everything, including KSP - but it changed nothing.

The only explanation I can come up with is that the MX-4 is somehow broken by default in this install, but I was able to use it once because I loaded a saved craft file that I had made in an earlier version.

The result, by the way, is this. Can anyone confirm they see the same?

The MX-1 shows how a reactor fresh out of the VAB should look. The MX.4 doesn't look anything like that. All the readouts are missing. You can start it, and it will begin producing heat internally, but it will never display any readouts. Without those, I cannot confirm whether or not it actually works as designed.

 

Edited December 12, 2015 by Streetwind

[Streetwind]

57 minutes ago, Psycho_zs said:

- with fully matched cooling the core does not heat up. If I had 5000kW of radiators on the ship, I would need to disable some of them to get to nominal temp.

 

Yeah, that's pretty much what I meant when I said that Reactor Heat Transf doesn't actually do anything yet aside from looking pretty. The reactor still transfers the full amout to the radiators at the moment. That's why it won't heat up with matched cooling, because you're producing and transferring away exactly the same number.

 

Edited December 12, 2015 by Streetwind

[Psycho_zs]

2 hours ago, Streetwind said:

Lithium frack yeah! 8) You can say what you want, but this is so much more badass than the hydrogen plume.

+1. Colossus is especially awesome.

It now looks a bit similar to hydrogen NTR exhaust when using RealPlume, although not as spread out.

[Streetwind]

Well, lithium and hydrogen do have similar spectra, so that's to be expected. Though I'm not sure that the NTR would excite the hydrogen atoms quite that much. All of the NERVA test videos I've seen showed a pretty much 100% colorless plume. Which is of course too boring for a game

 

By the way, bug report: VASIMR Isp does not scale with atmospheric pressure. Module VariableISPEngine must be overwriting it.

 

Edited December 12, 2015 by Streetwind

[Streetwind]

@Nertea, I put together a quick sheet illustrating the result of the mass ratio advantage of fuel tanks, since that relationship cannot be represented by a single number. It is here: https://dl.dropboxusercontent.com/u/44754370/massratio.xlsx - please excuse the rough nature of the sheet, I already prettified it a bit from the original mess that was what happens when I do stream-of-consciousness data analysis

It's meant to help pick a good value for lithium tanks. The current value of 1.5:1 is included, as are others all the way up to 2.0:1. You can even modify the numbers in the cells with borders to create additional scenarios with different numbers.

The idea is that you pick a value you like, and based on that number, I'll decide how to stat up the MPDTs. The higher the tank advantage, the lower the raw stats are going to end up. Ideally, building ships with MPDTs should feel reasonably different; if we're asking the player to go the extra step of looking at those engines differently from the others, then the difference should be marked so that the player has an epiphany moment of understanding and appreciating the differences. Personally, I am almost tempted to try 2.0:1

While we're at it: currently all three MPDTs are statted as tier 3 engines. Do you fancy moving one or more of them up or down, perhaps? Right now the Colossus comes one tech level later than the two smaller ones, but still one tech level earlier than the tier 4 VASIMRs. I could stat it up as a "tier 3.5", so that there is at least one lithium option not too far off of the other endgame engines.

[Nertea]

41 minutes ago, Streetwind said:

@Nertea, I put together a quick sheet illustrating the result of the mass ratio advantage of fuel tanks, since that relationship cannot be represented by a single number. It is here: https://dl.dropboxusercontent.com/u/44754370/massratio.xlsx - please excuse the rough nature of the sheet, I already prettified it a bit from the original mess that was what happens when I do stream-of-consciousness data analysis

It's meant to help pick a good value for lithium tanks. The current value of 1.5:1 is included, as are others all the way up to 2.0:1. You can even modify the numbers in the cells with borders to create additional scenarios with different numbers.

The idea is that you pick a value you like, and based on that number, I'll decide how to stat up the MPDTs. The higher the tank advantage, the lower the raw stats are going to end up. Ideally, building ships with MPDTs should feel reasonably different; if we're asking the player to go the extra step of looking at those engines differently from the others, then the difference should be marked so that the player has an epiphany moment of understanding and appreciating the differences. Personally, I am almost tempted to try 2.0:1

While we're at it: currently all three MPDTs are statted as tier 3 engines. Do you fancy moving one or more of them up or down, perhaps? Right now the Colossus comes one tech level later than the two smaller ones, but still one tech level earlier than the tier 4 VASIMRs. I could stat it up as a "tier 3.5", so that there is at least one lithium option not too far off of the other endgame engines.

Thanks for the spreadsheet, I'll have a look. I'm tempted to say 1.75 offhand, but I'll tweak the numbers.

Colossus could be tier 3.5 - it's supposed to be good.

4 hours ago, Streetwind said:

Well, lithium and hydrogen do have similar spectra, so that's to be expected. Though I'm not sure that the NTR would excite the hydrogen atoms quite that much. All of the NERVA test videos I've seen showed a pretty much 100% colorless plume. Which is of course too boring for a game

 

By the way, bug report: VASIMR Isp does not scale with atmospheric pressure. Module VariableISPEngine must be overwriting it.

 

Oh interesting. Yeah I don't think I ever really tested that, I'll have to fix it. Pretty nasty bug.

7 hours ago, Streetwind said:

 

Yeah, that's pretty much what I meant when I said that Reactor Heat Transf doesn't actually do anything yet aside from looking pretty. The reactor still transfers the full amout to the radiators at the moment. That's why it won't heat up with matched cooling, because you're producing and transferring away exactly the same number.

 

I couldn't get the throttling of the transfer to work, changing the value seemed to do nothing, so core temp now only scales the amount of power available to reactor consuming systems. Basically its heatTransfer = min (heat moved by radiators, (1- currentTemp/NominalTemp) * maximumCoreHeat). I do have an idea to fix this, but I'm not hopeful. If it doesn't work, I will need to get RoverDude to add some kind of override for 1.1 ;).

[Streetwind]

48 minutes ago, Nertea said:

I couldn't get the throttling of the transfer to work, changing the value seemed to do nothing, so core temp now only scales the amount of power available to reactor consuming systems. Basically its heatTransfer = min (heat moved by radiators, (1- currentTemp/NominalTemp) * maximumCoreHeat). I do have an idea to fix this, but I'm not hopeful. If it doesn't work, I will need to get RoverDude to add some kind of override for 1.1 ;).

Hmmm... well, good luck with that. If it doesn't work, maybe as a stopgap measure just scale Ec production directly with the power slider for the time being? The concept of nominal temperature becomes sort of redundant through that, but at least the player can properly control the amount of Ec and heat produced, and the amount of fuel consumed. It'll be functional at the same level it was in 1.0.4.

Good lord, this stock heat system stuff is such a neverending story, isn't it

 

Edited December 12, 2015 by Streetwind

[Captain Sierra]

3 hours ago, Streetwind said:

Good lord, this stock heat system stuff is such a neverending story, isn't it

(forgive any gross oversights as my coding knowledge is somewhat limited). At this point if it were my project, I'd do the reactor code in a vacuum. Design it to be entirely self contained and then only at the very beginning and end do I interface with the API. Theoretically reduces update breakage and also gives more control. All your own code, no API interfacing except for very initial input and final output. Does it "stay close to stock"? More or less. You can still output the final values into the stock thermal system. But it seems like all the confusion and frustration with stock heating would be so much better solved by circumventing it rather than trying to bend it to your will.

[Streetwind]

New engine parameter suggestions submitted to Nertea for perusal... those were the last open major item for NFP. If/when implemented, only minor things remain.

Just closed 4 open issues on github due to the changes worked out over the past week, leaving only one open.

 

Edited December 13, 2015 by Streetwind

[Captain Sierra]

4 hours ago, Streetwind said:

New engine parameter suggestions submitted to Nertea for perusal... those were the last open major item for NFP. If/when implemented, only minor things remain.

Just closed 4 open issues on github due to the changes worked out over the past week, leaving only one open.

 

Care to share aforementioned parameters? Would enjoy browsing through the restats to get an idea of how it will balance out. (forgive me if its already on the repo, Sierra does not know how to Github well)

[Streetwind]

I keep that particular spreadsheet mostly private because I've not had the best experiences with sharing things pre-finalization. It's a wee bit more complex than the little bit of mass ratio math from before, as you can imagine, and I am particularly loath to repeatedly address newcomer misunderstandings and re-explain the whole balance paradigm.

Also I'm on mobile with no dropbox access

PM me after the update and I will let you peek if you must.

[Nertea]

NFP Test 4: Normalized Li tank masses/costs/temps. Fixed VASIMR atmospheric Isp scaling. Implemented revisions of balance numbers for engines (let me know if I typoed or missed anything). This appears to be (mostly) a working 1.05 version, so please report any minor bugs so I can lock it down.

NFE Test 4: Fixed a few minor inconsistencies. Implemented a reactor fudging factor that should prevent reactor sticking at low temperature. Again, this is getting to the point where I think it's stable, so I guess you could report any minor bugs (please no "how about you change it to work this way") 

HC Test 1: Back from the dead! Now does not include heat pipe, heat exchangers, insulators (maybe later). All radiators have been re-specced as high temperature radiators more suitable for reactor operations than stock radiators (you will need less surface area/number). Includes universal, fixed, conformal and surface radiators. Functionality should be pretty close to 100%, but radiator heating animations need probably a small plugin to get them to work the way I want. 

Edited December 15, 2015 by Nertea

[Psycho_zs]

When nominal heat production = nominal radiator cooling, MX-1 core temp stops around 750K. When transferring in and out of time warp, radiator flow temporarily decreases which allows the core to heat up. So, switching timewarp on and off eventually results in overheat.

So, you didn't like my idea about heat loss?

[Streetwind]

Giving reactors a quick look over since it's pretty much my bedtime, so the more lengthy stuff like going over NF Propulsion will have to wait till tomorrow...

I like how it feels now. You can properly see how the power setting slider affects the reactor's operation and what happens in response to it; any player should be able to understand how reactors work just from poking at the UI. The temperature scaling is smooth, there is no flickering and everything works in a predictable fashion, exactly as it should be.

 

...Well, almost. I discovered these two issues:

- MX-4 still has the UI broken for me, as reported before... unfortunately nobody commented on whether or not they can see the same thing I looked at the part config and, apart from the order of lines, found no major differences to the MX-1... very odd.

- Reactors still don't stop creating energy out of thin air when deactivated. They'll cool down to ambient, but at that point, heat transfer to radiators hasn't scaled down to 0 yet. And thus, even though there is no more heat to transfer, the fully offline reactor remains stuck with a non-zero power production. This isn't just a visual issue on the UI either, but there is actual power production happening. Sorry Nertea but I'd class that as a major/critical bug

 

Also, something odd I noticed: the M-EXP reactor reaches its nominal temp of 650 K at a setting of 58 to 59 percent power... and keeps scaling up from there. At 100% power setting, it doesn't reach steady state until about 918 to 919 K. While that wasn't hot enough to do damage to the core, or interfered with the reactor's function in any way, I was nevertheless surprised. Not sure about the other reactors, this is the only one I had time to really look at.

 

Edited December 15, 2015 by Streetwind

[Psycho_zs]

When switching back to a craft with powered up reactor (was at balance) after spending some time on another craft including in time warp, lots of parts are overheated, reactor shutting down from overheat and core has some damage.

[Psycho_zs]

HC radiators:

Max temp seems low, 1900 vs 2500 stock

Something is wrong with Max cooling (at least as show in VAB). Example: XR-1500: 57kW, stock large radiator: 85077kW

I supposed, the feature of HC radiators would be higher temp difference, but VAB says 'Cools up to 4x part temp', just like stock radiators.

[Captain Sierra]

6 hours ago, Nertea said:

HC Test 1: Back from the dead! Now does not include heat pipe, heat exchangers, insulators (maybe later).

I take that "maybe later" to mean that the part-local heat management system is something that you're certainly willing to bring back, but still on the fence about. If I may add my $0.02, those of us who enjoyed that pack in its previous form can make due (some of us more willingly than others) with just the pretty radiators for now. I'd suggest holding off on bringing that back until the great 1.1 upheaval. No point fixing it now only to be broken again so soon.

[Nertea]

3 hours ago, Captain Sierra said:

I take that "maybe later" to mean that the part-local heat management system is something that you're certainly willing to bring back, but still on the fence about. If I may add my $0.02, those of us who enjoyed that pack in its previous form can make due (some of us more willingly than others) with just the pretty radiators for now. I'd suggest holding off on bringing that back until the great 1.1 upheaval. No point fixing it now only to be broken again so soon.

No it will probably never happen again. I'm kinda done with cooling and heating and stuff.

6 hours ago, Psycho_zs said:

HC radiators:

Max temp seems low, 1900 vs 2500 stock

Something is wrong with Max cooling (at least as show in VAB). Example: XR-1500: 57kW, stock large radiator: 85077kW

I supposed, the feature of HC radiators would be higher temp difference, but VAB says 'Cools up to 4x part temp', just like stock radiators.

Max cooling is probably something like emissivity*sigma*radiativearea*maxtemp^4. So with stock radiators it'll be crazy high because they're silly large and have silly maxtemps. As you note the maxTemp of the new radiators is a lot lower so that T^4 is really going to be lower. The key thing is the Max Transfer number, that's the max a radiator will ever take from a CoreHeat module. This is very low for stock radiators, compared to their overall ability to radiate, these are higher. The net effect will be that stock radiators will be a little better at radiating overall vessel heat (higher max radiation) but not be as effective at cooling cores. 

8 hours ago, Streetwind said:

Giving reactors a quick look over since it's pretty much my bedtime, so the more lengthy stuff like going over NF Propulsion will have to wait till tomorrow...

I like how it feels now. You can properly see how the power setting slider affects the reactor's operation and what happens in response to it; any player should be able to understand how reactors work just from poking at the UI. The temperature scaling is smooth, there is no flickering and everything works in a predictable fashion, exactly as it should be.

 

...Well, almost. I discovered these two issues:

- MX-4 still has the UI broken for me, as reported before... unfortunately nobody commented on whether or not they can see the same thing I looked at the part config and, apart from the order of lines, found no major differences to the MX-1... very odd.

- Reactors still don't stop creating energy out of thin air when deactivated. They'll cool down to ambient, but at that point, heat transfer to radiators hasn't scaled down to 0 yet. And thus, even though there is no more heat to transfer, the fully offline reactor remains stuck with a non-zero power production. This isn't just a visual issue on the UI either, but there is actual power production happening. Sorry Nertea but I'd class that as a major/critical bug

 

Also, something odd I noticed: the M-EXP reactor reaches its nominal temp of 650 K at a setting of 58 to 59 percent power... and keeps scaling up from there. At 100% power setting, it doesn't reach steady state until about 918 to 919 K. While that wasn't hot enough to do damage to the core, or interfered with the reactor's function in any way, I was nevertheless surprised. Not sure about the other reactors, this is the only one I had time to really look at.

 

This is probably the result of a bug that made the zero point for max temp a fixed number at 270K instead of the part skin temp, which it should be

The MX-4, if you look carefully you will find an incorrect cfg line.

7 hours ago, Psycho_zs said:

When switching back to a craft with powered up reactor (was at balance) after spending some time on another craft including in time warp, lots of parts are overheated, reactor shutting down from overheat and core has some damage.

I found a thing I didn't account for in smoothing transfers that caused significant dips in power production at high time warps and possible overheats (not enough power produced to run radiators for a few cycles). This should be fixed in the future but it would be nice to have solid repro steps.

[Streetwind]

5 hours ago, Captain Sierra said:

I take that "maybe later" to mean that the part-local heat management system is something that you're certainly willing to bring back, but still on the fence about. If I may add my $0.02, those of us who enjoyed that pack in its previous form can make due (some of us more willingly than others) with just the pretty radiators for now. I'd suggest holding off on bringing that back until the great 1.1 upheaval. No point fixing it now only to be broken again so soon.

Stock radiators now support part-local heat management, by the way. That was new in 1.0.5.

The fixed radiator panels are no longer passive, but active radiators just like the extending ones. But in contrast to them, they don't cool the whole ship - they only cool the part they are attached to, plus all parts attached to that part. If you look at their configs, they have an additional line in the heat radiator module that says something like "maxHops = 2".

You can manually add and modify this line for any radiator you want. So with just a simple MM config, you can set the foldable stock radiators to maxHops = 1, and they'll only cool whatever they're directly attached to. Voila, part-local heat management!

 

1 hour ago, Nertea said:

This is probably the result of a bug that made the zero point for max temp a fixed number at 270K instead of the part skin temp, which it should be

The MX-4, if you look carefully you will find an incorrect cfg line.

I'm glad you were able to narrow down these things so quickly!

I'll have another look at the MX-4 config, maybe I can spot the error myself when I'm not in a rush.

 

Edited December 16, 2015 by Streetwind

[Psycho_zs]

4 hours ago, Nertea said:

This should be fixed in the future but it would be nice to have solid repro steps.

Well... Have a craft in orbit with powered up reactor and matched cooling capacity at balance. Switch to any other craft outside physics bubble. Spend a couple of hours or days on it with time warp (in multiple bursts). I never used maximum timewarp steps, but gone over x1000. Switch back to original ship.

[ComatoseJedi]

It appears that a lot of hair has been lost in this matter and I quite frankly don't blame you all. I sense that most of you are burned out on this issue, but you guys just kept on chugging along. I've never seen so much devotion before in trying to make things right. You can't even get this kind of devotion on stuff in real life, much less a mod. All I have to say is: If I could buy you all a drink, I would and thank you for all you've done to keep this mod afloat.