Engine sizes and performace comparasion

[SRV Ron]

With one design for long range probes, I noticed quite a difference between using Skippers and Mainsails on first and second stages. Of the two designs, the skipper powered launch the second stage into orbit on the single capsule probe design with fuel to spare to reach Mum and place it into orbit. The same design as a three capsule with Mainsails, besides being difficult to control, needed the third stage to orbit. Of the two, Skippers were not powerful enough for the bigger payload but Mainsails were too inefficient. So a redesign was in order.

From a tutorial, I found how to cluster the smaller engines to get the power I needed for the larger capsule, In this case, four of those engines provided more power then the Skipper, but less then the mainsail. They were not only more fuel efficient, the overall rocket with the six cluster first stage attached to the second onion style, was a lot easier to control for the orbital turn. Fuel in the first stage lasted longer, no overheat problem, and the launch speed was just right for atmospheric travel. Result, three Kerbals reached a 180K orbit with the second stage still 1/2 full of fuel.

I sent them into Mum low orbit using the second stage and still had 641 x 784 units of fuel and oxider left. The three Kerbals still have a third stage and thruster tank to play with although no lander legs.

Total launch weight of the revised design using 28 of the smaller engines on launch then the four on second stage, 451 tons. In low Mum orbit, with the second stage and about 10% of its fuel left, 51.5 tons.

The challenge, designing something that can efficiently launch heavy tonnage into orbit without becoming so unwieldy that its launch weight becomes ridiculous or you seem to be just adding more fuel and rockets with very little gain. This sim does have a sweet spot in launch speed through its atmosphere and gravity. Find that as I did tonight and you will be amazed at the greatly improved results.

[arq]

I like mainsails for large vessels because they have over 2x the power of skippers. If you're TWR is ever far below 2 during launch (except for the final bit of circularizing orbit), you're prob better off with more engines, even if they're less efficient. Really, you want your TWR to be in the 2-3 range (higher later in the ascent).

For putting up large payloads, mainsails are a must. To put up a 400t payload the whole vessel is roughly 3000t at launch. That is 20 mainsails *just for a TWR of 1*. To get that thrust from skippers would require 45 of them (and the engines would weigh 50% more) . Now since you want a TWR around 1.7 on your first stage, you're looking at more like 34 mainsails. Doing that with skippers (or god forbid anything smaller) would be a partcount nightmare (over double the number of engines, plus the extra stuff to mount them).

Both engines have their place. Mainsails have lower ISP, but also higher TWR. So they keep partcount down, especially on your first stage (which is typically half the parts of my full vessel), and have similar efficiency to skippers given the lower weight. My general rule is that if I can get a high enough TWR with just a skipper then use it. If it isn't as high as I want, replace them with mainsails. When it comes to asparagus (actually I use onion) staging with large payloads, mainsails really are the answer. At the end of the day, I use mainsails much more than I use skippers.

As you said though, the smaller engines have higher yet ISPs, if you really want to get off on the cheap, you just need to cluster a gazillion aerospikes. Though this only works for small and medium sized vessels because of just how many it requires.

[SRV Ron]

On one probe design, I had to use the third stage outrigger rockets to complete orbit and reach Mun. Interestingly, their removal allowed me to achieve mun orbit, or go to a orbit inside of Eve on the second stage alone. Total range of that rocket was also improved since its launch weight to power available was in that ideal range.

An advantage of the smaller rocket cluster, more precise power placement, use of non vectoring engines on the booster section to get added stability on launch phase, weight savings, more efficiency. Disadvantage, higher part count, possibility of damage during staging should the ejection motors fail to kick the onion cluster out of the way, and frame rate dropping on launch.

[durk]

In low Mum orbit

Must resist making "yo mama is so fat" joke

[wilt57]

By the time you add the additional fuel tanks to feed the hungry mainsails and all the struts to hold those fuel tanks together, clustering doesn't add all that many additional parts.

[arq]

Yup, you have accurately addressed the advantages of smaller engines. Every tool has its place.

As for outboard engines decreasing range - I usually find that this is a result of adding sections with low TWR, as that can make your launch less efficient meaning you burn a lot of extra dV getting to orbit (sometimes more than you added with the extra tanks). Though there are other ways this can happen, too.

[sturmstiger]

... If you're TWR is ever far below 2 during launch (except for the final bit of circularizing orbit), you're prob better off with more engines, even if they're less efficient. Really, you want your TWR to be in the 2-3 range (higher later in the ascent).

Could you elaborate on why you want that high a TWR? As far as I know real world launch vehicles rarely have launch TWR > 1.5 (the exceptions are those derived from solid ICBMs). My KSP rockets tend to have launch TWR of around 1.3 for a good payload fraction (10%~15%). What are you optimizing by having a high TWR?

[KerbMav]

Whenever I built a first stage with a Mainsail surrounded by solid and liquid boosters - which drop off i.e. empty real fast, faster even then the solid boosters - once the boosters are all gone, I tend to have the problem of TWR < 0 (thank you Kerbal Enginieer) and therefore declining vertical speed, sometimes resulting in the rocket dropping out of the sky.

Firing up radial engines of the next stage only fixes this for a time, because they run out of fuel too and I am left with yet another remaining stage - often a single Skipper - burning fuel to not much result.

Asparagus staging is not working for me then for most it seems.

[m1xte]

Engine cluster and aspargus rockets are "cheap tricks".

Don't get addicted once the need to lift something stupid that weights 200t in to the orbit emerges. You can forget them both.

[maackey]

KerbMav: TWR cannot be less than zero, it is a ratio of two positive values. When it is below 1, you fall down, when it is above 1, you go up. When you get high enough, you can circularize with an engine with TWR < 1 because an orbit is just falling towards the planet and missing. You just have to get enough horizontal speed to "miss the planet" with your skipper (or whatever low TWR engines you're using).

m1xte: engine clustering and fuel crossfeed (asparagus or onion) are *extremely* useful *especially* for getting higher mass payloads into orbit. Engine clustering adds more parts, and fuel crossfeed adds more design complexity, but if you can afford either, they are much more efficient.

[nabeav]

Engine cluster and aspargus rockets are "cheap tricks".

Don't get addicted once the need to lift something stupid that weights 200t in to the orbit emerges. You can forget them both.

What? How is asparAgus staging a cheap trick? It does nothing but add efficiency. The reason we don't use it in the real world is because Earth's atmosphere is much different (taking into account the scale difference) than Kerbin's. Kerbin's atmosphere is much more dense, so methods such as asparagus staging are often used.

[NoMrBond]

Don't forget the Toroidal Aerospike, it can be nice on the outer boosters which you shed early/low instead of the LV-T30's because of its high ISP in the dense lower atmosphere (388)

Edited July 1, 2013 by NoMrBond

[AustralianFries]

I've never seen such a lack of avatars in one thread ever...

[arq]

Could you elaborate on why you want that high a TWR? As far as I know real world launch vehicles rarely have launch TWR > 1.5 (the exceptions are those derived from solid ICBMs). My KSP rockets tend to have launch TWR of around 1.3 for a good payload fraction (10%~15%). What are you optimizing by having a high TWR?

There was some analysis a long while back where it was decided that, based on KSP's drag model, the most efficient (delta-v) ascents are ones that follow terminal velocity all the way to orbit (or as far up as possible, because above 30km terminal velocity starts getting very high very fast), especially for the vertical ascent part before a gravity turn.

Terminal velocity means <drag> == <weight>, so in a uniform atmosphere you need TWR == 2. The atmosphere is not uniform (it decays exponentially), so actually you need to accelerate to keep terminal velocity. This means your TWR should be in the low 2's to get up in the smallest delta-v, and should increase to the mid 2's as you get around 15km. Lower TWRs will still get you to orbit, but nominally require slightly more delta-v. I try to keep my TWR at least 1.6 for every stage, except for circularization where 0.7 or sometimes lower can work.

Asparagus staging is not a cheap trick. It was not invented in KSP but existed long before (though rarely in practice because of issues with cross-feed). It is also very difficult to avoid with very heavy payloads (in my above example with a 400t payload, you need over 30 mainsails on the *first* stage just to get off the pad at an acceptable TWR - how many stages are you planning on stacking with conventional staging? the VAB is only so tall).

Clustering as a cheap trick? Ever looked at the back of the space shuttle? I count 5 engines. In KSP it's just a matter of wanting to get different mixes of thrust and ISP when we only have 10ish engines to choose from and can't custom-design each one. I don't do it because of part count issues and simplicity, but precisely nothing wrong with it.

[MR4Y]

Asparagus staging is not a cheap trick. It was not invented in KSP but existed long before (though rarely in practice because of issues with cross-feed). It is also very difficult to avoid with very heavy payloads (in my above example with a 400t payload, you need over 30 mainsails on the *first* stage just to get off the pad at an acceptable TWR - how many stages are you planning on stacking with conventional staging? the VAB is only so tall).

I wouldn't try to pass this one too further. What we have in real life is paralel staging. Asparagus staging is just paralel staging with fuel crossfeed added in.

[Seyvern]

I wouldn't try to pass this one too further. What we have in real life is paralel staging. Asparagus staging is just paralel staging with fuel crossfeed added in.

Up till now, we (the global we) have had parallel staging. However, Falcon 9H is going to have true asparagus staging, albeit with only one asparagus stage. So its not really a cheap trick.

[NoMrBond]

Sorry, the clustering bit just

Not sure why my brain did this

[numerobis]

The T30 has both better TWR and better Isp than the Skipper, so I usually skip the skipper and just add three T30s.

Of course, the Skipper has thrust vectoring, which is somewhat handy, but I find that a small number of strategically-placed T45 work well enough and still keep my overall engine TWR similar to the skipper's.

If you only need one engine, a mainsail lower stage followed by a skipper second stage followed by an upper stage works pretty nicely.

[KerbMav]

KerbMav: TWR cannot be less than zero, it is a ratio of two positive values. When it is below 1, you fall down, when it is above 1, you go up. When you get high enough, you can circularize with an engine with TWR < 1 because an orbit is just falling towards the planet and missing. You just have to get enough horizontal speed to "miss the planet" with your skipper (or whatever low TWR engines you're using).

That being the problem - at least with my early designs of heavy lifters - maybe I have them stored in the assembler addon still, I recently wiped my saved crafts from all older designs - not that I would do better with heavy payloads by now, I just concentrate on smaller landers now that my space station is up and running.

[m1xte]

m1xte: engine clustering and fuel crossfeed (asparagus or onion) are *extremely* useful *especially* for getting higher mass payloads into orbit. Engine clustering adds more parts, and fuel crossfeed adds more design complexity, but if you can afford either, they are much more efficient.

Aspargus rocket puts all stress to the central column, it is difficult to get enough control authority to aspargus rocket + lovely staging accidents.

Something stupid that weights 200t

Look mom no launch clamps .

Look mom can i actually control my 2kt rocket

http://imgur.com/a/0IxcP

Don't let aspargus staging hold you back just because its "efficient".

Move on people.

Edited July 2, 2013 by m1xte

[SRV Ron]

More experimenting with design efficiently. If one can get an ideal acceleration from the first stage, an underpowered second stage will achieve orbit with additional fuel to spare. The Redstone design had almost three times more fuel left by using a lighter less powerful engine. The Gemini design with a Poodle in place of the skipper in the second stage saw a two fold increase of fuel left.

Also, played with an Atlas design which is basically the Gemini first stage with the Redstone second stage using a smaller T-400 fuel tank and fitted with a thruster pack just for something to play with in orbit. Performance was good and like the real Atlas almost achieved 100K orbit. 10% of the second stage fuel was only needed for a 100+K orbit. 4.2 tons in orbit. A two step burn was needed on the Atlas first stage with the 10% burn from the second stage. Interestingly, the Gemini fell into a path that used a full burn to low orbit.

Edited July 3, 2013 by SRV Ron

[numerobis]

Aspargus rocket puts all stress to the central column

Not at all -- there is no central column! If you have the same starting TWR in every stage, then there's just the stress of the outermost stage having huge TWR when it's about to run out of fuel.

Your comment about control it is, indeed, something to worry about. I made a design that spiraled upward precisely to solve that issue: I had a bunch of non-vectoring stages spiralling upward, to make sure the later stages' vectoring engines had plenty of torque.

[Frederf]

I would be very interested to see scatter plots of various engine models (perhaps as a Compare-o-Tron feature of KSP itself) of Thrust vs Mass, Thrust vs Isp, Isp vs Mass, etc.

[Tiron]

I would be very interested to see scatter plots of various engine models (perhaps as a Compare-o-Tron feature of KSP itself) of Thrust vs Mass, Thrust vs Isp, Isp vs Mass, etc.

Well there's already a table with all the relevant statistics...

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

[m1xte]

Not at all -- there is no central column! If you have the same starting TWR in every stage, then there's just the stress of the outermost stage having huge TWR when it's about to run out of fuel.

Your comment about control it is, indeed, something to worry about. I made a design that spiraled upward precisely to solve that issue: I had a bunch of non-vectoring stages spiralling upward, to make sure the later stages' vectoring engines had plenty of torque.

Connection between payload and lifter is the weak link.

When stage is about to run out of fuel huge TWR of outer stage causes flexing in large rockets.

Flexing causes rocket to start spinning and/or to veer out of course.

Which causes need to steer the rocket back on course and all this happens at time of staging where huge TWR of the empty stage is about to turn to drag and dead mass.

Kaboom! Add more struts.

If 1/5 launches is success "efficiency" is pretty poor.