Finally turned to math to figure out why I hate the ion drive...

[ajburges]

I must confess. I have never found the ion drive terribly useful. It never seemed to offer much more dV than NERV engines and was more expensive for the craft. Since I have operational, recoverable, heavy lifters by the time I get ion engines, craft mass is a non-issue. There's also the additional issue that you can't refine more Xeon.

I started playing with numbers and derived the following equation to determine the asymptotic dV value for an engine:

lim dV as m[payload]->0 = ln( f[engine] * m[full tank] / (f[engine] * m[empty tank] + 9.81* m[engine] * (m[full tank] - m[empty tank]) * TWR) ) * Isp * 9.81

With this equation in hand we can plot different engine dV limits as a function of minimum TWR:

White: Poodle with orange tanks

Purple: NERV with mk1 tanks (note: mk0 is more mass efficent)

Red: Dawn engine with no EC

Green: Dawn engine with 6 OX-4* panels an engine to supply EC (conveniently .4 T/engine total)

Lo and behold the answer to my question. For a minimum TWR of .3, NERVs offer better asymptotic behavior! Since most my designs target .3 TWR for ejection burns, NERVs will offer better single stage dV limits.

Another interesting result of this exercise is the intercept between NERV and Poodle. A TWR of .6 matches up fairly well with the observations that NERVs can be dubious lander engines for the larger planetoids (Moho, Val, and Mun).

I found this exercise useful and thought I would share these findings.

[Cunjo Carl]

This style of analysis is pretty convenient, and it's nice to have an updated one for v.1.2. Thanks for the work!

Like you, I've found in practice that by the time I get enough energy collection on a Dawn it's not worth its weight. I think the way people typically get around the problem is by doing their burns in little spurts using (lighter) batteries to store just enough energy to move a bit, and then recollect the energy over a long period using with a less extensive solar panel array. Sounds tough! Also, once you add in your payload things get even harder for the little Dawn! For many people it's the ideal engine though. It can certainly be used to push some impressive extremes for deltaV... Here's a thing from Turbo Pumped that makes a nice case and point: https://www.youtube.com/watch?v=zolEcd2VzAY

Also, I think a big impetus for Dawns is that they're much smaller (so work well for probes). A nice analysis technique that shows this (from a much older version) can be found here:
http://imgur.com/a/OS6bk#5

I'm with you though. Heavy lifters are fun!

[Tweeker]

 

[KerikBalm]

a minimum TWR of 0.3 is quite high.

Also note, the fixed panels are not what you'd want to use. 

http://wiki.kerbalspaceprogram.com/wiki/Parts#Generators

The rotating but non retractable ones generate 32% more EC/ton.

Also, you may want to see this thread:

http://imgur.com/a/N1xNb#21

Nervs are heavy. If you want high deltaV at 0.3 TWR, but your payload is under 10 tons, you want the Ion engine. Methian did take the solar panel mass into account:

Quote

The curve labelled ION+SPs is the PB-ION with its mass increased by 100 kg, which is about the weight of the solar panels needed to fully satisfy the electricity demand at Kerbin distance. Note that Xenon tanks have wet-to-dry mass ratios of about 2.3 (so α about 1.3)

also note that if you plan multiple shorter maneuvers (like going from moon to moon around jool for example, changing polar to equatorial, landing on pol, etc), you can use batteries in place of more solar.

 

 

Edited February 18, 2017 by KerikBalm

[ajburges]

5 minutes ago, KerikBalm said:

a minimum TWR of 0.3 is quite high.

Also note, the fixed panels are not what you'd want to use. 

http://wiki.kerbalspaceprogram.com/wiki/Parts#Generators

The rotating but non retractable ones generate 32% more EC/ton.

Also, you may want to see this thread:

 

• .3 minimum TWR being too high for ions was one of the explicit conclusions.
• the OX-4* parts are the panels you mention.
• I didn't mention it, but one of the reasons I did that math is the web app does not use proper tank mass ratios; xeon canisters are all at least 43% dry mass! It portrays ion engines better than they actually are.

[Streetwind]

@ajburges - astutely observed. The Dawn engine does scale rather badly. But the reasons for this are not found in the engine itself or the power generation, suprisingly. While the engine itself is unusually heavy for its size, changing the weight doesn't have that much of an influence, and RL ion drives do require extensive power management electronics that could be seen as a reason or the weight.

No, the reason is in the tanks. When you use LF/O mix, the tanks have a mass ratio of 9. When you use pure LF for the nuclear rocket, you generally get a mass ratio of 8 (though the mk1 tanks have 9 and the tiny mk0 tanks have 10). But xenon tanks? Mass ratio of 2.27. As a result, you get diminishing returns almost immediately after the first tank you add. In roundabout ways, mass ratio is a multiplier on Isp that determiens how much dV you get - and when you just can't get the mass ratio up due to poor tanks, the engine's scaling is ruined, no matter how much Isp it has.

So as a result, right now the ion engine works best in situations where you only need a few tanks, with relatively low stage mass fractions... or, in other words, the ion engine works best at the lower end of its capabilities, with vessels that don't have all that much dV (as silly as that sounds). It's main draw is that it pulls off moderate amounts of dV with less mass and volume than other engine types, making ion-driven spacecraft cheaper to launch because the launcher itself can be smaller.

I've already brought this up as feedback to the devs over a year ago, with the suggestion to raise tank mass ratios while rebalancing the ion engine so that it doesn't scale out of proportion. The devs actually picked up on it and seemed positive on the idea, trying it out in a test build during 1.1 experimentals, and then earmarked it for the rocketry part revamp. Of course, then Porkjet left and most of the dev team with him, and the rocketry revamp was put indefinitely on hold... so I guess we're not gonna see any changes to this after all.

Well, for my part, I tweak my xenon tanks with MM to a value I consider acceptable without touching the engine stats themselves (mass ratios of about 3.0 to 3.5). A bit less dry mass on tanks means a little bit more TWR overall, and better scaling. But it doesn't change the fact that sometimes, the LV-N is still a better choice. You want to have niches for every engine type, after all, not just one engine that beats everything else.

Edited February 18, 2017 by Streetwind

[Delay]

I built a 2 ton satellite (to Eve) powered by a Dawn (of course, after circularization) At first I thought "Well, Xenon is pretty light, so dry mass won't be that different -> I need lots of fuel." Wrong!

I used the biggest one available; the one with 5250 units. After calculating dV, I found out that it gave me over 14,000 m/s: way too much. I decided to use two 700 unit tanks for around 3,000 m/s.

So no. In my opinion, the Dawn is pretty good when you ignore thrust to weight. Sure, it might take several minutes to perform a burn, but unless you want to land somewhere, efficiency is what really matters.

 

 

[Archgeek]

Yup, if you're okay with low TWR (just over .02 at worst), you can squeeze over 100km/s in just over 170 tonnes.  Hours-long stages, but just use 4x timewarp and KAC alarms set to pause the game, then all is good.  'Don't know the part count exactly, but my notes say that's 275 tanks of various sizes.

[GoSlash27]

I think the problem with the Dawn is that it's scaled all wrong for the KSP solar system. There simply isn't a need for the kind of DV numbers it provides, and by the time it can out- perform the NERV or chemical vacuum engines, it's t/w is too low to be convenient. Plus, it's way too expensive.

 It's a solution in search of a problem in stock KSP. I'm sure it fares better in RSS or similar modded installations.

Best,
-Slashy

[Archgeek]

2 hours ago, GoSlash27 said:

I think the problem with the Dawn is that it's scaled all wrong for the KSP solar system. There simply isn't a need for the kind of DV numbers it provides, and by the time it can out- perform the NERV or chemical vacuum engines, it's t/w is too low to be convenient.

What of high-energy transfers?  If you need something to get somewhere faster than normal, packing on more DV and taking an otherwise sub-optimal route can do the trick nicely.  Great for when a life-support-enabled mission is jeopardized and supplies need to be sent fast.

TWR's not much of a problem -- just use KAC and your aunt's gone and married Bob for some reason.

[GoSlash27]

Archgeek,

 I'm talking about stock KSP, specifically career.

Best,
-Slashy

[Archgeek]

10 hours ago, GoSlash27 said:

 I'm talking about stock KSP, specifically career.

Ah, rescue contract on a collision course then?  Say you take a contract to rescue a kerbal stuck in Duna orbit, send you rescue craft off that-a-way, and find out that sometime in the next few months, just before you'll get there, the poor clod's gonna slam into Ike?  There's less than one month left before impact, so, high-energy transfers to the rescue!  Just build an ion-powered beast (with some fuel cells for braking, at that distance) and send it ripping off to the aid of the imperiled kerbal, getting there in weeks instead of months.

Upshot being that unreasonably large delta-v can come in handy if there's ever an equally unreasonable time constraint in play.

Granted, the example in question would require you to've nearly o'ertopped the tech tree without having a functioning and ready craft already at Duna or Ike, even if only a powered probe that could at least bump the poor sod off their Ike-crossing course.  Though, that's just not as fun as a high-energy transfer in a specially built emergency rescue vehicle. 

[GoSlash27]

2 hours ago, Archgeek said:

Ah, rescue contract on a collision course then?  Say you take a contract to rescue a kerbal stuck in Duna orbit, send you rescue craft off that-a-way, and find out that sometime in the next few months, just before you'll get there, the poor clod's gonna slam into Ike?  There's less than one month left before impact, so, high-energy transfers to the rescue!  Just build an ion-powered beast (with some fuel cells for braking, at that distance) and send it ripping off to the aid of the imperiled kerbal, getting there in weeks instead of months.

Upshot being that unreasonably large delta-v can come in handy if there's ever an equally unreasonable time constraint in play.

Granted, the example in question would require you to've nearly o'ertopped the tech tree without having a functioning and ready craft already at Duna or Ike, even if only a powered probe that could at least bump the poor sod off their Ike-crossing course.  Though, that's just not as fun as a high-energy transfer in a specially built emergency rescue vehicle. 

Archgeek,

 I can't say I've ever gotten a rescue contract for a Kerbal on a collision course. *shrug* I'm not saying that excessively high DV is utterly useless... just sayin' I've never found a use for it in stock.

 Best,
-Slashy

[Archgeek]

21 minutes ago, GoSlash27 said:

 I can't say I've ever gotten a rescue contract for a Kerbal on a collision course. *shrug* I'm not saying that excessively high DV is utterly useless... just sayin' I've never found a use for it in stock.

Heheh, I've not either, but just have a looksee at one of the threads grousing about Ike causing craft loss.  Among the laments can be found rescue contracts that run into Ike and either get kicked into a solar orbit, into Duna, or just slam into Ike itself.

[wumpus]

18 hours ago, GoSlash27 said:

I think the problem with the Dawn is that it's scaled all wrong for the KSP solar system. There simply isn't a need for the kind of DV numbers it provides, and by the time it can out- perform the NERV or chemical vacuum engines, it's t/w is too low to be convenient. Plus, it's way too expensive.

 It's a solution in search of a problem in stock KSP. I'm sure it fares better in RSS or similar modded installations.

Best,
-Slashy

The other huge problem is that the T/W is unrealistically high, although this is to deal with KSP's time acceleration system rather than the Kerbol system.  Your best bet for ion engines would more likely be RSS with percipia (real gravity overhaul mod) added (assuming it can handle time acceleration with an "accelerating" craft, which is pretty much what it has to do with n-body gravity).

The big real life problem (that I'm reasonably sure won't be modeled in KSP) is that I've heard that solar panels have issues with the Van Allan belt (by heard, I mean somebody mentioned it in the forum).  I've only tracked down an Apollo-era paper about it, but NASA seems to spend a ton of delta-v making sure ion engines (and their solar panels) don't have to deal with the Van Allen belts.  Since the delta-v needed to get *anywhere* in the Solar System (especially at ion engine speeds) isn't much more than the delta-v needed to get past the Van Allen belts, this is pretty much the nail in coffin for using ion engines for anything other than sending probes to multiple places in the Solar System.

[Kertech]

eith near future you get capacitors, they weigh a lot less than a battery, so get each one to be under your maximum battery charge capacity, then discharge them one by one whilst doing a burn, makes cheap, light ion probes that can operate at full thrust!! 

[Cunjo Carl]

9 hours ago, Kertech said:

eith near future you get capacitors, they weigh a lot less than a battery, so get each one to be under your maximum battery charge capacity, then discharge them one by one whilst doing a burn, makes cheap, light ion probes that can operate at full thrust!! 

I like the idea, but capacitors are heavy! Even the really good ones only store ~1/3 the energy of the same weight of cheapo lead-acid car batteries. What they lack in lasting energy though they make up for with an insane ability to push sudden power. They're a very kerbal power source.

[Streetwind]

2 hours ago, Cunjo Carl said:

I like the idea, but capacitors are heavy! Even the really good ones only store ~1/3 the energy of the same weight of cheapo lead-acid car batteries. What they lack in lasting energy though they make up for with an insane ability to push sudden power. They're a very kerbal power source.

Since batteries in KSP don't really work like IRL batteries, you won't really find much realism in NFT's capacitors either. You'll instead find them to be much, much lighter per EC stored.

(The main balancing factor is that they intentionally do not automatically dispense their EC. You have to manually request a discharge, and you better have a large regular battery available to receive the burst!)

[Jeb-head-mug kerman]

Ions are useful for small probes and landers, it's true that they have little power and you can't get more xenon. But if you have solar panels and a LOT of xenon you can find them ok.

[foamyesque]

On 2017-02-19 at 5:26 PM, GoSlash27 said:

Archgeek,

 I can't say I've ever gotten a rescue contract for a Kerbal on a collision course. *shrug* I'm not saying that excessively high DV is utterly useless... just sayin' I've never found a use for it in stock.

 Best,
-Slashy

I like using 'em for scanning satellites, which want to be in polar orbits, which means you're either executing plane change manouvers or capture burns. The ion engine is small and lightweight, and doesn't require a bulky fuel tank to go with it either; it's the perfect solution for shoving satellites around, and can even muster reasonable TWR numbers in the role (not great, but enough to, e.g., circularize from a capture orbit on Eve in a single burn). A single NERV would outmass the entire system by itself, and chemical engines start struggling to hit the dV targets for major bodies without your mass blowing out.

Edited February 21, 2017 by foamyesque

[DerekL1963]

5 hours ago, foamyesque said:

I like using 'em for scanning satellites, which want to be in polar orbits, which means you're either executing plane change manouvers or capture burns.

I keep hearing this...  and I have to wonder if I'm the only one who carries out the plane change right after entering the target SOI... where they're pretty cheap in terms of d/v.

[Cunjo Carl]

Ohhhh.... Thanks for the explanation, @Streetwind, that makes a lot more sense. Just as a side note for interest sake, it's actually pretty tricky to get capacitors to do that IRL. Superconductors, on the other hand, work exactly that way!

I like that for all the different options we have, folk are shooting for excess in different departments and all getting where they're going. Excess dV, excess TWR (or in my case, excess cost). Hm... Smacks of a well balanced game!

Along those lines, I just saw this impressive mission from @Jetski which made nice use of ion and LFO landers alike. 1 Huge battery + 2 ions made a quite convenient looking little ion lander.
http://forum.kerbalspaceprogram.com/index.php?/topic/154147-grand-tour-no-nukes/

 

 

1 hour ago, DerekL1963 said:

I keep hearing this...  and I have to wonder if I'm the only one who carries out the plane change right after entering the target SOI... where they're pretty cheap in terms of d/v.

Nah, I'm right there with you, I typically do a pair of small burns a couple weeks away to put myself on whichever plane and keep it that way. I can't imagine it's the most efficient, but it's easy to eyeball and certainly convenient. On the other hand, if you've got the dV to plane-change in orbit why not use it?

 

[foamyesque]

2 hours ago, DerekL1963 said:

I keep hearing this...  and I have to wonder if I'm the only one who carries out the plane change right after entering the target SOI... where they're pretty cheap in terms of d/v.

 

I said "or capture burn" for a reason (though, I suppose, more accurately, it's actually a circularization burn). Either you wind up executing your plane change in low orbit -- where it's damned expensive -- or you perform the burn necessary to go from a highly elliptical orbit into a low circular one, which is also a pretty big dV sink. Circularizing with aerobraking takes too damn long and has a much higher chance of something going wrong.

[DerekL1963]

14 minutes ago, foamyesque said:

I said "or capture burn" for a reason (though, I suppose, more accurately, it's actually a circularization burn). Either you wind up executing your plane change in low orbit -- where it's damned expensive -- or you perform the burn necessary to go from a highly elliptical orbit into a low circular one, which is also a pretty big dV sink.

You missed my point foamy - those aren't the only options, only the most expensive ones.   The further out you perform your plane change burn, the cheaper it gets - and doing it just inside the SOI is pretty cheap.   If you're a better pilot than I am, you can do it even cheaper as part of your mid-course correction burn.   The only way to "end up" executing plane changes using the methods you describe it to skip the earlier options.

[foamyesque]

Just now, DerekL1963 said:

You missed my point foamy - those aren't the only options, only the most expensive ones.   The further out you perform your plane change burn, the cheaper it gets - and doing it just inside the SOI is pretty cheap.   If you're a better pilot than I am, you can do it even cheaper as part of your mid-course correction burn.   The only way to "end up" executing plane changes using the methods you describe it to skip the earlier options.

 

What part of "or" are you not understanding here? Sure, you can perform your plane change at a high apoapsis (or, hell, even outside of the SOI if you know what you're about), but then you need to circularize, because a "barely captured elliptical polar orbit" is not in fact one in which a scanning satellite will operate.