100km LKO in only 2800m/s dV -- Ongoing Experiments

[Geschosskopf]

So I've been dinking around with trying to figure the optimal launch profile from Kerbin and thus the minimum dV required to reach an orbit of 100km. I'm by no means done with this but here are some preliminary results.

EXPERIMENTS

1st Round

I built a big rocket with 3184m/s (in vacuum) in the lifter part of it (it had other stages on top). The lifter portion was a central Mainsail core with 2 pairs of asparagus boosters using Skippers. I then launched this rocket repeatedly using MJ's ascent guidance using the following settings;

* 65% curve shape in all tests (because this shape always worked best for me with FAR)

* start turn at 1km above launchpad (best get started early with this curve)

* varied the altitude to stop the turn from 70km down to 35km.

With each turn altitude, I did 2 launches and got the same result both times so I figure this is consistent.

Results:

70km turn stop required 3233m/s (measured on the vacuum side of the table for all results)+

50km turn stop required 3129m/s

40km turn stop required 3071m/s

35km turn stop required 3050m/s

+ I only read the vacuum side because that's what I care about once it's up.

Conclusion:

It's best to level out at 35km, which is just above the last noticeable blue in the pressure indicator.

2nd Round

I'm in the habit of "overlapping" my stages (it cuts down on pesky debris). By this I mean the last part of the lifter does the 1st part of the transfer burn so it ends up well away from LKO, then the actual transfer stage does the 1st part of the landing so it will be sure to crash. Thus, the ship in Round 1 above didn't have quite enough left in the lifter for my tastes, so I added some fuel to the 1st set of asparagus booster, keeping the same engines of a central Mainsail surrounded by 4 Skippers. This had 3438m/s (vac) in the lifter and an initial TWR of 1.33. I then launched this rocket using the 35km turn stop altitude obtained in round 1.

Result: 3092m/s required to reach 100km, 346m/s left in the tank at 100km.

I then swapped out the Skippers on the 1st (bigger) asparagus boosters for Mainsails, keeping everything else the same. This dropped the lifter dV down to 3307m/s (vac) but raised the TWR to 1.84. Again, launched with the same ascent profile (1km start, 65% curve, 35km stop).

Result: 2789m/s required to reach 100km, 518m/s left in the tank at 100km.

Conclusion:

For a given flight profile, a higher TWR trumps a higher dV total, at least up to a point. The 3-Mainsail/2-Skipper version got some mach effects shortly before the Mainsail boosters dropped, which I'm sure is inefficient. Plus, Kerbals don't like long periods of high Gs. Thus, it's probably not a good idea to overpower your rockets. FAR didn't like to exceed about 1.5 Gs for most of the launch, which is why I tried to keep my initial TWRs rather low.

BOTTOM LINE

An ascent profile that has you level off about 35km will save you fuel. So will higher TWRs, even if you have less dV to start with. The main point, however, is that the "new standard" of 3400m/s to LKO is way too high. It should be closer to 3000m/s, maybe even less. I can do it in 2800m/s if I try hard enough.

Future Study

The obvious next step is to vary the shape of the curve and also the altitude to start turning. I suspect (as it was in FAR) that a curve that starts slowly and finishes quickly (that is, > 50%) will prove the best. The idea is to get through the thicker part of the atmosphere quickly, which probably explains wihy having a higher initial TWR also helps, then drop those engines for more fuel efficiency once you've got some altitude and speed.

NOTE: I"m sure most vets have figured this out already. I say this for the benefit of the players, especially those coming in and seeing folks saying it takes 3400m/s to get to LKO.

Here's a pic of my ridiculous test rocket in both 1-Mainsail / 4 Skipper side-by-side with the 3-Mainsail / 2-Skipper version. See for yourself.

[Starwhip]

Seconding the modified profile. I level out at around 25-30k. At that altitude, there is basically 0 drag (And also 0 air, bye-bye, air-hoggers! ). I'm also noticing the effects of high-TWR. The new atmosphere is much more forgiving to high-thrust rockets, in fact, it is beneficial to escape the 10k mark as quickly as you can. I stay at 80 degrees until 5k, and by 11k I pitch to 45 degrees. Stay at 45 until apoapsis breaks 60k, then pitch to 0 degrees.

[Core21]

I seem to not be able to reproduce the same dV to orbit numbers as you with the given ascend profile.

How is your staging set up?

I noticed mechjeb displyway too low dV numbers when you set all engines to fire at once and then drop some of the boosters in flight, like in your asparagus setup.

Also what are the mechjeb ascend stats reporting?

[marach]

You're seeing the effect of the changed isp system. Remember fuel rate is now constant and different engine are more effective at different altitudes. You'll need to keep an eye on atmospheric dv now since iirc mainsails are designed for launch, skippers are sustainers and the poodle should be vacuum...

Also starwhip you should be aiming for gravity turns, so smooth turns with no notches following the prograde marker for best efficiency

[magnemoe]

I seem to not be able to reproduce the same dV to orbit numbers as you with the given ascend profile.

How is your staging set up?

I noticed mechjeb displyway too low dV numbers when you set all engines to fire at once and then drop some of the boosters in flight, like in your asparagus setup.

Also what are the mechjeb ascend stats reporting?

Mechjeb does not understand drop tanks or asparagus in 1.0 it calculates it as if you did not drop anything before you go upper stage.

[Falkenherz]

My initial tests in career mode come to the very blunt result that a high TWR is always better, if your rocket parts can sustain the stress.

I first experimented with just one "Hammer" solid fuel booster and different TWR settings. The maximum height I could achieve was 42km and only with a TWR of 3 onwards. Everything lower topped out at about 30km.

Later, I did countless "tourist in suborbit" missions, with hammer solid fuel boosters S1+S2 plus triple Mk1 command pods and modified the trust. Thus playing with TWRs of 1.6 to 3 and just going up at 10 degrees off vertical, I found that I could reach a maximum height of 72km from a TWR from 2.2 onwards only.

What I read here in the forum and my theory about it is that on lower atmosphere, drag is reduced by breaking the sonic barrier, which in turn allows the rocket to go farther. A high TWR helps breaking the sonic barrier asap. Again, just my current theory.

[KerikBalm]

If you are using only the vacuum dV number, then of course you are going to get different numbers with different engines, even if you keep the same TWR.

The LV-909 or the LV-N will suffer terribly in the first part of the ascent... whereas the mammoth or the aerospike will not.

The engines that have a smaller Isp range will give you better dV numbers than those that have a high dV range.

Isp changes with altitude, so your vacuum dV numbers do not represent the true dV spent....

Still, its nice to know we're now losing ~ 500 m/s to aero and grav drag to get to orbit (approximately, considering one orbits at about 2,300 m/s in low orbit)

[Wolf Baginski]

I have only made launches so far in science mode, but this does fit with what I observe.

Done right, you can get out the atmosphere with less fuel than v0.90 but you need very fine timing on the burn to get the circular orbit.

The single nose chute will land on hard ground with a service bay and heatshield.

Battery management changes a lot with the need to keep close to retrograde alignment for re-entry. It seems crazy but radio depends on batteries. It may be safer to not fit a radio.

Burnout with a periapsis of about 65km can give you Science without a need for radio and a certain landing somewhere, eventually as long as you have been careful with the battery.

[GoSlash27]

I seem to not be able to reproduce the same dV to orbit numbers as you with the given ascend profile.

How is your staging set up?

I noticed mechjeb displyway too low dV numbers when you set all engines to fire at once and then drop some of the boosters in flight, like in your asparagus setup.

Also what are the mechjeb ascend stats reporting?

This is the important part; a limit is only so useful for planning purposes if people can't achieve it.

MechJeb can be easily confused by multiple stages, so it may not be giving you an accurate reading (even though it's consistent with itself).

Also, the 3400 m/sec figure is assuming vacuum Isp throughout the flight, which would yield a DV budget somewhat higher than what's actually used.

I'll try and see if I can reproduce your results with my sounding rocket.

Best,

-Slashy

Edited April 29, 2015 by GoSlash27

[Streetwind]

Besides the things already mentioned re: atmospheric Isp and thrust relationship, there's something else to consider:

In the past it was easy to give a single number like 4500 m/s for the launch to Kerbin orbit because craft shape literally didn't matter. But now it does. That means differently shaped rockets are going to require different amounts of dV to orbit, even if they fly the exact same launch profile with the exact same TWR.

Blanket figures are a lot less reliable now than they were in the past, and dV maps will 8and should) probably err towards the higher end of the range.

[Geschosskopf]

I seem to not be able to reproduce the same dV to orbit numbers as you with the given ascend profile.

How is your staging set up?

The launch clamps and all 5 lift engines are on the 1st stage, then the decouplers for the bigger boosters, then the decouplers for the smaller boosters, then the inline decoupler for the core. Fuel lines run from the big boosters to the small boosters, and from the small boosters to the core.

I noticed mechjeb displyway too low dV numbers when you set all engines to fire at once and then drop some of the boosters in flight, like in your asparagus setup. Also what are the mechjeb ascend stats reporting?

For the purposes of training to figure out the most efficient launch profile, it doesn't matter how MJ calculates its numbers as long as it does its calculations the same way each time. This allows making relative comparisons between different launch profiles for the same rocket, which was the whole reason for running these tests. It's not the absolute numbers that matter, it's the trends.

This is the important part; a limit is only so useful for planning purposes if people can't achieve it.

MechJeb can be easily confused by multiple stages, so it may not be giving you an accurate reading (even though it's consistent with itself).

Also, the 3400 m/sec figure is assuming vacuum Isp throughout the flight, which would yield a DV budget somewhat higher than what's actually used.

Right, just like the old 4500m/s standard was based on the vacuum Isp. Which is a logical thing given that for the vast bulk of any space trip, it's the vacuum Isp that matters, starting with what's left in the tank immediately after launch. The goal, therefore, is to fly the launch in such a way that you have as much vacuum dV after launch as possible for whatever stage you're using for the next phase of the mission. Thus, unlike a spaceplane, with a rocket you really don't care about atmospheric Isp except to avoid the obvious pitfalls like trying to launch with LV-Ns or (as of 1.0) 909s. For most other engines, there's really not a lot of difference between atmospheric and vacuum performance anyway.

Is using vacuum Isp and dV numbers for planning purposes absolutely accurate? No. However, it does produce useful rules of thumb because the whole process is internally consistent and easily understood. Now, if you want absolute accuracy, and to carefully design and perfectly fly a rocket right on your calculated path, you certainly can. And you just might get more efficiency that way. But do you get enough to make any real difference for gameplay purposes, in terms of cost? Probably not. Plus, you spend all that time worrying about just getting off the ground instead of actually going places. Maybe that's your thing and if so, rock on. But most folks just want to kick the tires and light the fires, so simple rules of thumb that reliably and accurately enough approximate the detailed calculations are good enough.

Besides the things already mentioned re: atmospheric Isp and thrust relationship, there's something else to consider:

In the past it was easy to give a single number like 4500 m/s for the launch to Kerbin orbit because craft shape literally didn't matter. But now it does. That means differently shaped rockets are going to require different amounts of dV to orbit, even if they fly the exact same launch profile with the exact same TWR.

Blanket figures are a lot less reliable now than they were in the past, and dV maps will 8and should) probably err towards the higher end of the range.

This is quite true. If you try to fly a pancake, it will burn more fuel than a needle of the same mass and theoretical dV on board. I suppose I could have put in a caveat about that, but OTOH I assume that most folks would be aware of this given the emphasis placed on the new aerodynamics. Still, however, the same processes work on all rockets during launch. Thus, all rockets should benefit in the same way from the same sorts of ascent path refinements, although the blunter rockets won't get as much benefit as the sleeker ones.

Another design issue folks now need to be aware of in stock is the relative position of the CoM and the CoL, from which stability and the ability to control the rocket during launch derive. As with an arrow, the closer to the top the CoM, and the closer to the bottom the CoL, the better the rocket will fly. This is, however, not nearly such an issue as it is in NEAR/FAR. Classic KSP-style asparagus staging of multiple huge boosters, as shown in my pic above, is still quite possible, whereas it really isn't in NEAR/FAR. Also, if you need to move the CoL further aft in a stock rocket, you get a lot more bang for a smaller number of smaller tail fins than you do in NEAR/FAR. The upshot of this is that nowadays, as you build your rocket from the top down, you need to check the CoM/CoL placement of each stage that will be flying in the atmosphere and fix that prior to adding more stages below and to the sides, and checking them as well taking the pre-existing stages into account.

I suppose some folks will complain that the new air is a half-measure because it doesn't doom the classic KSP pancake lifter, it just imposes a few constraints that can be dealt with. To which I say, if somebody can make a pancake fly, and is willing to pay the price, good for them if that's what they consider fun. Besides, the way the new tech tree is laid out, combined with the need to level up multiple Kerbals at once, means that mission payloads grow faster than single stacks of the then-available rockets can support. Thus, you need to be able to do some asparagus, especially considering that you now have to buy several top-tier nodes to get all the parts needed to use 3.75m rockets. And that's all there is in stock. To avoid pancakes and apsaragus after that, you need a mod with 5m rockets.

But all the above just goes to show that how you can spiral off into all sorts of issues if you want to over-analyze the seemingly simple question of "how much dV do I need to get off the ground?" I'm just saying that even with all the new factors involved, you can still give this a simple answer, provided you make some underlying assumptions.

[A_name]

Great analysis OP, please keep us updated on the results of your further tests.

[Francesco]

I have only made launches so far in science mode, but this does fit with what I observe.

Done right, you can get out the atmosphere with less fuel than v0.90 but you need very fine timing on the burn to get the circular orbit.

The single nose chute will land on hard ground with a service bay and heatshield.

Battery management changes a lot with the need to keep close to retrograde alignment for re-entry. It seems crazy but radio depends on batteries. It may be safer to not fit a radio.

Burnout with a periapsis of about 65km can give you Science without a need for radio and a certain landing somewhere, eventually as long as you have been careful with the battery.

Battery management? Radio??

Could you explain this?

[Gfurst]

Good to know, I love efficient ascent profiles. But as it happens playing on career mode now and mostly without SAS and fighting off wobliness, its quite hard to keep on course, let alone maintain the curve.

Getting off initially with lower TWR helps sure, is there any way to maintain course instead of mechjeb?

[GoSlash27]

But all the above just goes to show that how you can spiral off into all sorts of issues if you want to over-analyze the seemingly simple question of "how much dV do I need to get off the ground?" I'm just saying that even with all the new factors involved, you can still give this a simple answer, provided you make some underlying assumptions.

Exactly so, which is why we're using the value that we are.

If you build an "average" rocket with the vacuum Isp figure in the VAB, 3,400 m/sec should be about what is required to make LKO. The absolute minimum expended during the launch is a nice figure to know (especially for learning how to operate efficiently), but less useful for planning purposes.

Incidentally, I was unable to shave off any waste by running your profile. I still wound up in the 3,400m/sec ballpark (although the *true* figure probably lined up closely with yours).

I was running a constant curve right off the pad, reducing the t/w as a function of 2sin(pitch) until I reached a hard minimum of 1G.

Best,

-Slashy

[Radam]

There really is a profile like that:

just about 3000 dv to orbit.

Not sure how to get to 2800 tho. More thrust I guess.

[rkman]

If you use Mechjeb, why not use its delta-V 'recorder' function?

Hadn't seen that you are using the dv recorder.

So minimizing d-v to LKO iis just a matter of making the ascent trajectory very flat and go as fast as possible, drag losses don't matter much.

Even on planets with an atmosphere all that matters is minimizing gravity loss, which does not seem to be what one would expect.

Edited April 30, 2015 by rkman

[SkyRender]

I noticed that the heating from aerodynamic forces also cuts out around 32 to 35km, which helps since otherwise you hit the "death wall" (that being the speed at which parts exposed to the airstream start exploding; 1200m/s is the usual speed for that at any altitude). Once you're past that point, it's smooth sailing.

[A_name]

Seconding the modified profile. I level out at around 25-30k. At that altitude, there is basically 0 drag (And also 0 air, bye-bye, air-hoggers! ). I'm also noticing the effects of high-TWR. The new atmosphere is much more forgiving to high-thrust rockets, in fact, it is beneficial to escape the 10k mark as quickly as you can. I stay at 80 degrees until 5k, and by 11k I pitch to 45 degrees. Stay at 45 until apoapsis breaks 60k, then pitch to 0 degrees.

When you say as quickly as you can how fast is that? I'm having trouble keeping my rockets from catching on fire during ascent.

[Renegrade]

So minimizing d-v to LKO iis just a matter of making the ascent trajectory very flat and go as fast as possible, drag losses don't matter much.

Even on planets with an atmosphere all that matters is minimizing gravity loss, which does not seem to be what one would expect.

That's exactly what one should expect.

Typical losses for spacecraft reaching Earth orbit are on the order of 1200-1500m/s for gravity 'drag', and 40-156m/s for aero drag.

The Saturn V, one of the lowest aero drag launchers (it's kinda slow in the low atmosphere, plus it's large mass and small cross-section, gives it a high grav drag, low aero drag profile) only loses 0.4% (yes, less than half a percent) to air.

The most correct modelling of the atmosphere for Kerbin would be to set it as a hard vacuum~

Using infinite TWR rockets in a vacuum, one could theoretically reach a 200km orbit on earth for only about 8,031m/s total delta-v (ignoring rotation cuz I'm lazy). Same situation on Kerbin (again ignoring rotation) is 2,606m/s for a 100km orbit.

[GoSlash27]

All,

KER has not been updated to reflect the new Isp of the 1.0 engines. The numbers that it generates are invalid.

^ Not correct. KER works fine in 1.0.

Best,

-Slashy

Edited May 1, 2015 by GoSlash27

[Renegrade]

All,

KER has not been updated to reflect the new Isp of the 1.0 engines. The numbers that it generates are invalid.

Best,

-Slashy

Doesn't it get the Isp from the parts themselves? O.o

The only thing I see as amiss is that it and VOID slightly disagree, but that could very well be that the 9.82 nonsense (I'd call it 'idiocy' but I'm trying to be nicer these days) was fixed for 1.0. The middle-click info for KER is showing the correct values.

For instance:

Command pod mk1 (0.84t)

T800 fuel tank (4.5t - 4t fuel)

T30 "Reliant" (1.25t, 300s vac isp)

6.59t total, 4t fuel, 300s vac isp.

Windows calculator: 2,747.5158115089721650995629816004

My own tool, using 9.82 for Standard Gravity: 2751.256/vac

My own tool, retooled for CORRECT g₀ value: 2747.516/vac

KER: 2,751

VOID: 2,748

...close enough?

Has that g₀ issue officially been fixed? (Maybe I should get off my lazy duff and read the WHOLE changelog finally?)

[rkman]

That's exactly what one should expect.

Typical losses for spacecraft reaching Earth orbit are on the order of 1200-1500m/s for gravity 'drag', and 40-156m/s for aero drag.

The Saturn V, one of the lowest aero drag launchers (it's kinda slow in the low atmosphere, plus it's large mass and small cross-section, gives it a high grav drag, low aero drag profile) only loses 0.4% (yes, less than half a percent) to air.

That's because they do stay below terminal velocity, contrary to what we do in KSP.

The most correct modelling of the atmosphere for Kerbin would be to set it as a hard vacuum~

And how is that exactly what one should expect?

[Eric S]

KER has not been updated to reflect the new Isp of the 1.0 engines. The numbers that it generates are invalid.

KER is fine, but if you want to see your thrust/ISP on the launchpad, you have to toggle the atmospheric/vacuum control.

[GoSlash27]

KER is fine, but if you want to see your thrust/ISP on the launchpad, you have to toggle the atmospheric/vacuum control.

Oh, okay. I picked that one uphere, but it's clearly erroneous.

Disregard my last...

Sorry for the confusion!

-Slashy