Comprehensive Space Engine Efficiency By-Usage Analysis

[Venusgate]

Something that i've done a lot of tinkering with over the 1.1k+ hours playing KSP is "What engine and type of fuel usage is best for a type of craft?"

And to preface, I'm not talking about air-breathing engines, and I'm not going to talk about ion engines, except by exception.

Now, for example, if you have a craft that shuttles cargo/fuel/passengers between a Mun base and an orbital Mun space station, the distance your craft travels on one tank of gas is not very far. If it's below 20 tons, A terrier would probably be better on efficiency over a NERVA, but maybe a few mono engines and the added weight of a mono tank replacing a LFO tank would be more efficient? What If it's 40 tons? What If you're hauling 30 tons of LF?

Here I call on my fellow engineers to share their experience not of preference and ease of use (although that's certainly a big part of craft design) but on experience and analysis of efficiency of engines in outer space (that is, no launch stage) and where they are most efficient based on size of vessel, utility of vessel, and distance for vessel to travel.

----
Experience based conclusions:
"Nuke engine is the ONLY practical engine in space unless you are below 10t. " (panzer1b)

 

Edited December 2, 2015 by Venusgate
New data

[Temstar]

Quote

if you have a craft that shuttles cargo/fuel/passengers between a Mun base and an orbital Mun space station

Basically any time you see the words "shuttle between" and "reusable" and "deep space only" the answer is NTR engine.

For single use lightweight spacecraft then chemical engines become viable. There's a chart somewhere for this that shows what engine is best to use at a given TWR (ie the % of your craft that's engine determines which engine is best for most delta-V).

But for crafts that always stay in vacuum and you will refuel and reuse over and over again the answer is always the nuke engine, because its ISP is THAT good.

[CalculusWarrior]

Exactly, the NTR is the best engine for carrying large* payloads in space applications. For lifting things off Kerbin, you generally want the highest-ISP engine with as much thrust as is required to give you a TWR of around 1.5. For landers on airless bodies, your best bet is generally those engines which have high fuel efficiency and decent thrust, enough to give you a sufficient TWR for landing, anyways.

Niche engines like the aerospike (ships which spend a lot of time in-atmosphere) and the tiny engines (probes, obviously) are used on those more rare applications, and as such are not quite as sufficient for transporting large payloads.

 

*as a rule of thumb, payload of 5 or more tonnes

[panzer1b]

ill say right off the bat, that although i do not really like it, the NTR (nuke) engine is the ONLY practical engine in space unless you are below 10t.  Unless you are either extremely impatient, or have some very niche role where TWR needs to be high to accomplish your mission, a nuke or even a cluster of nukes will always outperform anything LFO or monoprop based.  Statistically the terrier (or even 48-7s) is better when vehicle mass becomes extremely low (say under 10t), but for anything heavier then 10t a nuke will near always get you more raw dV even if you remove 2.5t of fuel to make room for the 3t nuke (compared to a .5t terrier).  800 vs 350 is a major deal, ofc ions still put both to shame with 4200 ISP, but ions at least have their share of issues, terrible thrust (requires utter spamming for anything but a probe), require lot of batteries/solar panels, and well, they are just not useable with very large craft as youd need 100s of em.

Even if you look at the engine being rather heavy at 3t, it is still a known fact that nukes are the one and only practical interplanetary or space engine, and heck, they are so good that i even have a large amount of vessels that use nukes to land on other planets.  Anything Mun or less gravity is perfectly doable with a single nuke pushing ~20t, and duna can be done with less provided you have wings for actual lift and dont need to land vertically.

This is one thing about KSP that annoys me, since well, there really arent any options if all you care about if pure efficiency/dV.  Myself, i do use othyer engines when im designing for fun/looks and dont care that the craft is inefficient (made less problematic with IRSU and the seemingly endless supplies of fuel), but whenever i want maximum dV with minimum mass, its nuke or nothing.

I know this is a bit off topic, but i wish that stock game came with a larger ion engine (say 1.25m, with 10x scaling in stats, so 20kn thrust, and 10 times the fuel useage/electric needs).  Also, a different nuke would be neat, perhaps a 2.5m one with at least 200kN thrust, and ofc heavier, and either a 0.6m one that is smaller and ~20kN thrust, and perhaps a half as light and 30kN thrust one but shorter then what we get now.  Why the bloody nukes have to be so long and in my opinion ugly (well not ugly but just doesnt look right with my designs) is beyond me.  If we have to use nukes and only nukes for efficiency, then at least give us larger and smaller ones for both massive ships and smaller vessels.

Edited December 2, 2015 by panzer1b

[Venusgate]

So would you say the same 10t mark would hold if you were transporting between, say, Bop surface to bop orbit?

[Temstar]

3 minutes ago, Venusgate said:

So would you say the same 10t mark would hold if you were transporting between, say, Bop surface to bop orbit?

Yes definitely. Chemical engine only matter if TWR is an issue. Bop's gravity is so low that TWR is completely irrelevant.

Even with the Mun the gravity is low enough that nuke powered craft's TWR is a non-issue. TWR is only an issue when you're dealing with world's like Tylo. Tylo's gravity is strong enough that you need a lot of thrust to safely land and take off, and here chemical engines start to look good again because of their much higher TWR.

 

[Venusgate]

12 minutes ago, Temstar said:

Yes definitely. Chemical engine only matter if TWR is an issue. Bop's gravity is so low that TWR is completely irrelevant.

Even with the Mun the gravity is low enough that nuke powered craft's TWR is a non-issue. TWR is only an issue when you're dealing with world's like Tylo. Tylo's gravity is strong enough that you need a lot of thrust to safely land and take off, and here chemical engines start to look good again because of their much higher TWR.

 

So if you have a 10t payload and only need .5t in what chemical engines can provide, the 4+tons of a nuke is still better? Let's see some math, sir.

[Temstar]

Suppose your rocket can boost a 10 ton spacecraft into orbit. The goal is to maximize the delta-V of this craft.

Let's say out of this 10 ton payload, 2.5 ton is mission hardware, eg the reason why you're going into space in the first place, so you can't change this part.
You start with a nuke engine, which is 3 tons, this leave you with 4.5 tons to play with. This happens to fit two Mk1 jet fuel tanks perfectly. These two tanks weights 4.5 ton, of which 4 ton is the fuel.

So the dry weight of your craft is 2.5 + 3 + 0.5, or 6 tons. Plug 10 ton wet mass, 6 ton dry mass and 800s ISP into a delta-V calculator and you get 4007.59m/s of delta-V

 

Now suppose instead of a nuke engine we use a LV-909 Terrier instead. It has same vac thrust as a nuke but only weights 0.5 ton, it has a lot lower ISP though at 345s.

Because the engine is now 2.5 tons lighter, we free up 2.5 tons of payload capacity on our launch vehicle. Naturally since we want delta-V we devote all that 2.5 ton to more fuel.

KSP fuel tanks have a 9:1 wet/dry mass ratio. So a hypothetical 2.5 ton fuel tank will have a dry mass of 0.278 tons. Good thing bipropellent has the same combined density as jet fuel as this makes calculation easy.

So now, our chemical engined spacecraft has a dry mass of 2.5 + 0.5 + 0.5 + 0.278 tons, or 3.778 tons. We plug 10 tons wet mass, 3.778 tons dry mass and 345s ISP into delta-V calculator again and we get 3283.27m/s of delta-V.

So as you can see, despite having much more of the craft being fuel, the chemical engined craft has much lower delta-V than the nuke powered version.

[Teilnehmer]

1 hour ago, Temstar said:

There's a chart somewhere for this that shows what engine is best to use at a given TWR

Here it is: Meithan’s Optimal Engine Charts
(does not include Vector yet)

[Francois424]

3 hours ago, panzer1b said:

ill say right off the bat, that although i do not really like it, the NTR (nuke) engine is the ONLY practical engine in space unless you are below 10t.  Unless you are either extremely impatient, or have some very niche role where TWR needs to be high to accomplish your mission, a nuke or even a cluster of nukes will always outperform anything LFO or monoprop based.  Statistically the terrier (or even 48-7s) is better when vehicle mass becomes extremely low (say under 10t), but for anything heavier then 10t a nuke will near always get you more raw dV even if you remove 2.5t of fuel to make room for the 3t nuke (compared to a .5t terrier).  800 vs 350 is a major deal, ofc ions still put both to shame with 4200 ISP, but ions at least have their share of issues, terrible thrust (requires utter spamming for anything but a probe), require lot of batteries/solar panels, and well, they are just not useable with very large craft as youd need 100s of em.

Even if you look at the engine being rather heavy at 3t, it is still a known fact that nukes are the one and only practical interplanetary or space engine, and heck, they are so good that i even have a large amount of vessels that use nukes to land on other planets.  Anything Mun or less gravity is perfectly doable with a single nuke pushing ~20t, and duna can be done with less provided you have wings for actual lift and dont need to land vertically.

This is one thing about KSP that annoys me, since well, there really arent any options if all you care about if pure efficiency/dV.  Myself, i do use othyer engines when im designing for fun/looks and dont care that the craft is inefficient (made less problematic with IRSU and the seemingly endless supplies of fuel), but whenever i want maximum dV with minimum mass, its nuke or nothing.

I know this is a bit off topic, but i wish that stock game came with a larger ion engine (say 1.25m, with 10x scaling in stats, so 20kn thrust, and 10 times the fuel useage/electric needs).  Also, a different nuke would be neat, perhaps a 2.5m one with at least 200kN thrust, and ofc heavier, and either a 0.6m one that is smaller and ~20kN thrust, and perhaps a half as light and 30kN thrust one but shorter then what we get now.  Why the bloody nukes have to be so long and in my opinion ugly (well not ugly but just doesnt look right with my designs) is beyond me.  If we have to use nukes and only nukes for efficiency, then at least give us larger and smaller ones for both massive ships and smaller vessels.

I agree we are missing engines in-between what we have right now.

A few months ago when everyone was discussing the Nerv's someone had posted that given the choice of a chemical rocket of 150thr/600ISP and Nukes of 60thr/800ISP it would really make him think of which one to use,... I still agree to this day.  Ion Engines have a great ISP but are FAR to inconvenient to use with their pitiful 2 thr (I agree we need a 20/100/250 thr variants).  Nuke is good but again we are missing a 250 thr variant for bigger ships.   Not only because that would make for better engines, but rather to cut down on parts by a lot.   Let's face it, I like my motherships to have 2k thrust, and with current Ions(1000!!!) or Nervas (33) it's ridiculous.  Of course there are mods to get the parts and you can then mod them some to make them just as you like, but stock is missing out on key things.  Could use a large RTG but that's for another thread as well and I think Necrobones has made one already.

As for the actual topic of the thread, there is very few instances where I would consider anything else than a nuke engine.  When I build a small lander, I always try out the engines, and unless it is really, really small, the Nuke is always the better choice.  Obviously Tylo/Eve and other heavy gravity bodies, and also Moho (because of the usual short burn window I get to decelerate into orbit) I prefer a more powerful engine (Vector is my new favorite).  An in-between chemical engine between 450-600 ISP and 100-150 thr would really make me use something else than the Nerva here too.

 

 

5 hours ago, Venusgate said:

Now, for example, if you have a craft that shuttles cargo/fuel/passengers between a Mun base and an orbital Mun space station, the distance your craft travels on one tank of gas is not very far. If it's below 20 tons, A terrier would probably be better on efficiency over a NERVA, but maybe a few mono engines and the added weight of a mono tank replacing a LFO tank would be more efficient? What If it's 40 tons? What If you're hauling 30 tons of LF?

In the provided example, just for the sake of quick rendez-vous and burns, I would most likely use those radially attached Mk55 Thuds (like 4 of them), or a Vector if a big miner-ship/heavy cargo.  Nerva get annoying when using on anything Moon and over as far as gravity goes... It's definitely doable, but much more manageable with Thuds or something like that.

 

 

Edited December 2, 2015 by Francois424

[Geschosskopf]

Long ago I used nukes pretty much exclusively for any stage that would stay forever in vacuum.  However, the recent rebalances to engines over the last few updates, the addition of new engines, not to mention budget-consciousness in career mode, have altered my opinion and I use nukes a lot less than before.

In general, these days I use chemical interplanetary transfer stages wherever possible and you'd be surprised at how good some of them can be, even above the so-call "10 ton payload" limit.  Rhinos are quite good for this, often at considerably less overall vehicle cost than nukes.  In most cases, I now only use nukes for crew return vehicles and things that will do a lot of traveling at the destination, like visiting numerous moons of Jool.  But getting it out there to begin with, especially if it's not coming home?  Or if it doesn't need to do much maneuvering at the target,?  LFO, baby.

The reason for this is that transfer burns out from Kerbin need to be 5 minutes or less.  Any longer and you curve around Kerbin so much that you waste a lot of fuel pointing crossways to your current vector, have to start at a higher orbit so you don't dive into the atmosphere, etc.  Either that or spread the burn out over multiple orbits, which is rather time-consuming and reduces precision.  And that's a real problem if, like me, you typically send out large flotillas to other planets instead of just 1 ship per window so can't devote too much time to any one of them.

So here's the thing.  Burn time is a function of TWR and nukes have pathetic thrust.  So the only way to reduce burn time is to use a lot of them, which greatly increases cost and part count, plus takes a big bite out of the fuel efficiency you're using nukes for to begin with.  Really, to make the most of a nuke's ISP, you should only ever use 1 of them no matter how heavy the ship.  2 at most.  Certainly not 6, 8, or even more (which I was often guilty of in my younger days).  It's way cheaper and simpler just to use 1 Rhino instead and get the necessary TWR and shorter burn times automatically.

The bottom line is, to me an interplanetary transfer burn is the last step in the launch, just pushing of the Ap to the desired altitude, not the beginning of some new phase of the flight.  Thus, I regard transfer stages just like SRBs, things to get their simple job done as quickly as possible and then thrown away.  The mission doesn't start until the payload reaches the target.  So,. just as you don't use nukes for the atmospheric part of the ascent, why should you use them for the vacuum part?

[Foxster]

I'm not sure that the suggested 10t rule for the use of nukes is so cut and dry. You need a lot of mass with nukes and you have to haul them into space before they are useful; that takes a lot of fuel/dV.

Most of my interplanetary craft are disposable, so including lifting to orbit it can be better overall to take a hit on the transfer stage to save mass lifted to orbit.    

Also, if there is the need for a lander at the end of the journey then some Terriers or Thuds can be more useful than nukes and will still get you all the way home fine too. 

[Geher]

For interplanetary transfer burns with lots of payload I mostly use the Rhino. I don't want to use nukes, because I hate long burntimes. And what I even hate more is multiple burns

Edited December 2, 2015 by Geher

[Temstar]

If cost of interplanetary transfer is an issue the answer is still nukes. Because yes the transfer stage will be bigger, however you simply reuse the transfer stage by refueling it LKO with SSTO tanker spaceplances and/or refuel at target planet from ISRU bases.Fuel is cheap, throwing engines away is not. Remember also launch from Kerbin itself can be done with a dry transfer stage which can then be fueled in LKO with SSTOs, thus cutting down cost of launch vehicles again.

Yes nukes have bad TWR and long burns can be tricky, but it still doesn't mean you need to burn away from prograde - it's a simple matter of starting your ejection burn earlier so that by the time you reach your target velocity your ejection vector matches the phase angle for the interplanetary transfer. Your orbit will be "wider" from Kerbin than a short burn transfer with high TWR, but on an interplanetary scale that distance is negligible.

It's not like nuke transfer stages need burns that last 30 minutes or something, it's only slightly longer.

[Venusgate]

3 hours ago, Temstar said:

Suppose your rocket can boost a 10 ton spacecraft into orbit. The goal is to maximize the delta-V of this craft.

Let's say out of this 10 ton payload, 2.5 ton is mission hardware, eg the reason why you're going into space in the first place, so you can't change this part.
You start with a nuke engine, which is 3 tons, this leave you with 4.5 tons to play with. This happens to fit two Mk1 jet fuel tanks perfectly. These two tanks weights 4.5 ton, of which 4 ton is the fuel.

So the dry weight of your craft is 2.5 + 3 + 0.5, or 6 tons. Plug 10 ton wet mass, 6 ton dry mass and 800s ISP into a delta-V calculator and you get 4007.59m/s of delta-V

 

Now suppose instead of a nuke engine we use a LV-909 Terrier instead. It has same vac thrust as a nuke but only weights 0.5 ton, it has a lot lower ISP though at 345s.

Because the engine is now 2.5 tons lighter, we free up 2.5 tons of payload capacity on our launch vehicle. Naturally since we want delta-V we devote all that 2.5 ton to more fuel.

KSP fuel tanks have a 9:1 wet/dry mass ratio. So a hypothetical 2.5 ton fuel tank will have a dry mass of 0.278 tons. Good thing bipropellent has the same combined density as jet fuel as this makes calculation easy.

So now, our chemical engined spacecraft has a dry mass of 2.5 + 0.5 + 0.5 + 0.278 tons, or 3.778 tons. We plug 10 tons wet mass, 3.778 tons dry mass and 345s ISP into delta-V calculator again and we get 3283.27m/s of delta-V.

So as you can see, despite having much more of the craft being fuel, the chemical engined craft has much lower delta-V than the nuke powered version.

I meant .5tons over 10tons of payload as a few ants + LFO to go... what... 200dV? Vs 10t payload + Nuke(s) + enough LF to go 200dV.

[problemecium]

I don't like absolutes (sith happens) but panzer1b has the right idea based on my experiences. For small probes, single-occupant landers, etc. that only weigh a few tons (or less than one), I find stuff like the 24-77 "Spark" (and of course ion engines) preferable, but for anything with significant heft the LV-N's awesome Isp makes it superior. Even conservatively sized things at 10 tons or so tend to be better off with LV-Ns, particularly in 1.0 in which they don't require Oxidizer.

[Streetwind]

55 minutes ago, Venusgate said:

I meant .5tons over 10tons of payload as a few ants + LFO to go... what... 200dV? Vs 10t payload + Nuke(s) + enough LF to go 200dV.

10 tons payload + LV-N + NCS Adapter (silly, yes, but it's the smallest liquid fuel tank): 13.5t wet, 13.1t dry, 236 m/s dV, 7.546 Bop-relative TWR, 10,320 funds and 2 parts (drive section only)

10 tons payload + FL-T200 + 6 cubic octagonal struts + 6 Ants: 11.251t wet, 10.251t dry, 287.5 m/s dV, 1.811 Bop-relative TWR, 1,031 funds and 13 parts (drive section only). Using just a FL-T100 does not exceed 150 m/s dV.

Does that help?

Edited December 2, 2015 by Streetwind

[Venusgate]

9 minutes ago, Streetwind said:

10 tons payload + LV-N + NCS Adapter (silly, yes, but it's the smallest liquid fuel tank): 13.5t wet, 13.1t dry, 236 m/s dV, 7.546 Bop-relative TWR, 10,320 funds and 2 parts (drive section only)

10 tons payload + FL-T200 + 6 cubic octagonal struts + 6 Ants: 11.251t wet, 10.251t dry, 287.5 m/s dV, 1.811 Bop-relative TWR, 1,031 funds and 13 parts (drive section only). Using just a FL-T100 does not exceed 150 m/s dV.

Does that help?

So in this case, the ants are, in fact, more efficient?

[Streetwind]

Define "efficient"? They consume more fuel by mass to do the same job, which means more refueling effort; they have much worse TWR; but, the entire drive section costs significantly less. There's advantages and disadvantages - take your pick.

Oh, and just for fun:

10 tons payload + 2x PB-X50R + 3x Z-4k batteries + 3x Z-1k batteries + 6x cubic octagonal struts + 6 Dawns:
12.396t wet, 12.316t dry, 266.7 m/s dV, 1.644 Bop-relative TWR, 66,636 funds and 20 parts. For when you absolutely, positively must prove that your space program is overfunded

[Venusgate]

42 minutes ago, Streetwind said:

Define "efficient"? They consume more fuel by mass to do the same job, which means more refueling effort; they have much worse TWR; but, the entire drive section costs significantly less. There's advantages and disadvantages - take your pick.

Efficient in this case being more dV for a given job. That's the simplest criterion.

Talking outside of that, if we use Spiders instead, that can cut more weight, and if you use procedural parts, you can shave down to the exact dV you need, more or less. As well as a tighter, lighter craft is going to be easier to maneuver with RCS, and without those honkin' NERVAs hangin out, designing a landing gear system would be less challenging. But again, those are all semantics outside of the hope of mt intent.

[Venusgate]

5 hours ago, Teilnehmer said:

Here it is: Meithan’s Optimal Engine Charts
(does not include Vector yet)

Simply marvelous!

EDIT: So playing around with this it looks like at least for low mass planets, payload is not as much a factor as dV cutoff is. Around 300dVit where the efficiency jumps from chem engines to NERVAs.

Edited December 2, 2015 by Venusgate
Research!

[Temstar]

The point is 400m/s is not really a useful amount. You can't even reach KSO from LKO with that delta-V. Yeah you can land on Gilly a few times, but if you want to get back to Eve you still need a transfer stage, and if you need to use a transfer stage you need to be able to dock with it and your RCS system is already powerful enough for landing and reaching orbit:


For orbital manoeuvring you need a bit of TWR, else burn time become unmanageable. So saying "well I can get heaps of delta-V with one ant engine and 100 tons of fuel" is a moot point - a spacecraft like that is unusable anyway. For practical application a deep space spacecraft need 1000+m/s of delta-V to go anywhere interesting and you need to be able to complete that burn in a burn time measured in minutes not hours. And pretty much in all such cases nuke is your best bit.

Of course I can think of cases outside this:
if you want to lower a probe to low, circular kerbol orbit for example then having burn time measured in hours is fine, and plentiful sunlight makes ion practial
If you want to land on Tylo you need a lot of TWR, proper big chemical engines are needed here
If you're making an small orbital runabout that only need to move between stations, then your delta-V requirement is so low that a small chemical engine will do
If you want to land and take off from Gilly to meet up with a transfer stage, your RCS blocks will suffice

For everything else, use nukes.
 

[Streetwind]

53 minutes ago, Venusgate said:

Efficient in this case being more dV for a given job. That's the simplest criterion.

Ah, but a given job implies a given dV, and payload mass is not a factor. In other words, hauling 1 ton from low Kerbin orbit to the Mun is going to cost you ~860 m/s, and hauling 100 tons to the Mun is also going to cost you ~860 m/s. As a result, doing the same job with more dV than what you actually need is less efficient, not more efficient.

If you sized both the chemical and the nuclear vehicle with procedural parts to exactly the specification of going 200 m/s with 10 tons of payload, the LV-N is still going to be "more efficient" in terms of reaction mass expended (eyeballed, it needs about 320 kg as opposed to about 600 kg for the Ants). But at the same time, it is "less efficient" in terms of what money you invested, and it made the launcher necessary to haul it out of Kerbin's gravity well and all the way to Bop larger and more expensive as well.

If you want a scenario where the Ants actually expend less reaction mass in addition to costing less, you need a smaller payload so that the total vessel mass is dominated more by the dry mass of the LV-N. The one thing that works out in the Ants' favor is that they have so much less dry mass, which means that in order to maximize the advantage, you want to minimize the influence of masses which do not belong to the drive sections - in other words, the payload. This smoothly results in what everyone intuitively understands already: Ants are best used for small probes, while the LV-N excels at hauling mass cargo.

Edited December 2, 2015 by Streetwind

[technion]

I don't understand how everyone is still fitting nukes. Since the rebalancing, putting one on either side of a large grey tank is no longer a great idea - it gets dv on par with a poodle.

The mk2 jet fuel tank looks and handles stupidly under a round command module, and a series of mk1 tanks doesn't scale very well.

[Streetwind]

- They don't use large grey tanks, because those are bad for LF only
- There are mk2 to 1.25m and mk2 to 2.5m adapters
- Most LV-N tugs I've seen use mk3 tanks, actually (and yes, there are adapters to 2.5m)

Also, mods.

Edited December 2, 2015 by Streetwind