Just how much better is a gravity turn than direct ascent?

[bewing]

Yes, there are a bunch of oldtimers who are going to moan and say "not this issue again!!?" Tough. Each new group of newbies needs to be exposed to the argument. If you don't want to see it, don't read any further.

But I hope several of you will try to show me the error of my ways.  @foamyesque & @Spricigo?

So, I'm in the direct ascent camp. The gravity turn zealots make wild assumptions about how wonderfully efficient it is to do it their way. I say the gains are minimal at best, and it depends on whether you are calculating "efficiency" on the price of a kg to Munar or Minmus orbit, rather than deltaV.

So let's try a competition. It only works if we all launch the same upper stage. So here's the one we are going to all use: http://www.virtualrealitytoursllc.com/pix/scan_sat.craft

Here's a pic. Note the price = 16720 kredits.

The rules are: you can add anything to the upper stage that you want to make it fly your way. You can try as many times as you like to be as efficient as you can. You can use MechJeb to fly it for you. Put the upper stage in (approximately) a 10 x 11.5 km orbit around your destination celestial body and report your total launch cost, the time it took to reach the destination, your remaining fuel, and the price per kg to reach that orbit.

However, you may not remove anything from the common upper stage. Only additions allowed.

Now, this is supposed to be a test of gravity turns. So SSTO spaceplanes are really not in the competition. But if any of you want to launch the upper stage as cargo in an SSTO and report the numbers, that would still be interesting.

 

And I made it easy for you. I launched mine by eye. I didn't use any tools or maneuver nodes or anything. You can beat me by a bunch if I do that, right??! I even gave you a break on the time. I started a new game at Day 1, 0:00 and waited several hours for a launch window.

Here's my launchable version of the rocket (note: grand total price = 30690 kredits):

 

So my total cost to launch the upper stage was 30690 - 16720 = 13970 kredits.  Proof that I got there -- and my mass in orbit = 7610 kg. Total flight time = 8 hours, 7 mins.

 

 

With 675 fuel remaining. Cost per kg in LMO was 13970 / 7610 = 1.84 kredits per kg.

 

And then I also did Minmus for fun. The numbers are almost identical. 1.85 kredits per kg. Total flight time (including waiting for window) = 6 days, 4 hr, 36 mins.

Edited August 16, 2016 by bewing

[foamyesque]

That's nice. I did it in under a third of that, though. Including the cost of the liquid fuel I used in the payload -- which you, I notice, did not.

401 funds in LF + a 3100 funds launch stage / 5.83t in Munar orbit = 0.6 funds per kg.

 

 

So basically, @bewing, the answer is: Lots.

 

EDIT:

It's also important to distinguish between a direct ascent -- that is to say, going somewhere without pausing in a parking orbit or to dock with something -- and a vertical ascent. You keep calling the latter the former. A direct ascent as conventionally defined is more efficient deltaV-wise than a parking orbit profile, but it's still a perfectly ordinary gravity turn that just raises the apoapsis to the desired transfer altitude in one go and doesn't bother with the circularization part. It gets benefits from Oberth and from not needing to execute that burn whilst retaining all the savings of not fighting gravity.

 

Going upright, well, you're fighting gravity the whole way.

 

EDIT2:

 

Managed to lower it with a better flight path and a tweak to the launch stage. Turns out I didn't need steerable fins.

Launcher cost, 2800. LF cost, 377 funds. Mass to Mun orbit, 5.98t.

 

Funds per kilo: 0.53.

 

I think I'll call it there for the evening.

Edited August 16, 2016 by foamyesque

[Superfluous J]

It's important to know WHY @foamyesque was able to do this with a single SRB, while @bewing was not: TWR of less than 1 will KILL you going straight up, but is very doable when most of your thrust is sideways. So, the second craft was able to ditch the high-TWR but low-efficiency SRBs earlier and rely on the super-efficient NERV engine for a lot more of the trip.

It wouldn't have been quite so lopsided if the original ship didn't have a NERV engine, and instead had an engine with a higher TWR. It'd still have been worse, just not 3 times worse. Also, using the payload's fuel in the test really muddies the water, making any real analysis difficult at best. That's why most of the time when we discuss launch profiles, we get a payload into LKO without using any of its fuel.

[Spricigo]

Disclaimer: since @bewing addressed his OP at me this post is mostly to reply to his arguments.  Because of that may look that I disregard something in the previous replies and the post may sound a bit harsh.  But it's  not my intention to disrespect anyone's opnion in any way. 

 

First of all : I'm not interested to prove you wrong.  Just what you said in the other thread don't convince me. As said there I feel it may be true for a few cases but hardly something that I wold rely on for every single launch. This challenge really will not change this.  It gives me no incentive to prove you wrong,  nor I feel challenged to put a random vessel in orbit around the Mun.

That said I really like to discuss gameplay,  it's a nice mental exercise and a very good opportunity to improve some mental skills that can be used elsewhere. Particularly there is a lot to learn from other people's approach. 

 

OK now about the discussion between vertical ascent Vs gravity turn:

Quote

I say the gains are minimal at best, and it depends on whether you are calculating "efficiency" on the price of a kg to Munar or Minmus orbit, rather than deltaV.

Let's be clear: assessment of efficiency base in deltaV used or deltaV left in orbit it's just silly. Just don't pull the strawman. 

Quote

... competition. It only works if we all launch the same upper stage. So here's the one... 

No.  The upper stage it's not what mater,  it's the payload. 

 You can't use half of this fuel and claim to be more efficient than someone that just noticed half of the fuel it's unnecessary and removed it before launch.  Neither he can claim to be more efficient than the guy that noticed the payload it's just the scanner satellite and brought it with a  tweaked ComSat_LX. 

Quote

Now, this is supposed to be a test of gravity turns. 

And that is why I find the challenge so uninteresting.  Gravity turn it's already tested and true.  Also we just scratch the subject,  once in orbit you can refuel or use a previous launched tug what will make single use in general and SRBs in particular even less competitive. 

 

Quote

So SSTO spaceplanes are really not in the competition.

Well,  not all SSTO are spaceplanes and not all recoverable vehicles are SSTOs.  Maybe  a comparison between SSTO rockets Vs Falcon style wold yield better arguments against the efficiency of perfect gravity turns. 

 

 

Finally I'm thinking about a challenge for some time and this one give me some insight to improve a bit my original idea. Thank you. 

[Spricigo]

3 hours ago, foamyesque said:

Going upright, well, you're fighting gravity the whole way.

That is the whole point. No matter what you do gravity will be there doing its thing. Instead of trying to overcome gravity the idea it's to flight in a such way that the effect of gravity will not hinder you. Once you have enough orbital speed you will not be falling toward kerbin instead you will be falling around kerbin. 

It's kind like not swing against the stream pushing you away from the beach but getting out the stream in the first place. 

[Plusck]

assuming the voice of an old-timer, though that qualification is a bit of a stretch...

"not this issue again!" 

However, if each new group of newbies needs to be exposed to the argument, they need a number of descriptions of why a vertical ascent can sometimes appear to be almost as efficient.

In any given orbit, it's a elementary fact that the higher your destination Ap, the faster you go at Pe and the less time you spend there.

A perfectly circular LKO at 70km has an orbital velocity of 2295 m/s. Losses to drag and gravity means that most people manage to get there by spending about 1000 m/s more.
To get to Minmus, you need to increase that velocity to 3224 m/s - an additional 1000 m/s (almost).

However, if you were to start on the surface and accelerate perfectly horizontally, you would only need a velocity of 3409 m/s to attain an Ap at Minmus. And you would spend a vanishingly small amount of time going "horizontally" - within a few minutes you would be well out of the atmosphere and almost vertical to Kerbin's surface.
In other words, with a highly eliptical orbit, you spend most of the time on a heading that is as near vertical as makes no difference.
Conceptually, therefore, it is easy to see how the gravitational pull on a vessel on a highly elliptical path is, for most of the long climb out of the gravity well, almost identical to a purely vertical ascent. The only real difference is for those short seconds to cover the 600km or so needed to get "past" Kerbin itself.

You can put it into mathematical terms to get the same result. The classic equation is v = sq rt (GM(2/r-1/a)). Where GM=constant for Kerbin's gravity, r=current radius, a=semi-major axis.
From this you can see that orbital dynamics don't care what shape your orbit is, at a given distance r from the planetary body, you will be going at velocity v for a given semi-major axis a. And the higher the destination Ap, the less of a diffference in "a" between different low Kerbin orbits: your exact altitude becomes largely irrelevant.
Therefore, if you launch vertically up, you're already at r=600km on an "orbit" that takes you through the centre of Kerbin. With "a" being half the altitude of Minmus, that gives a launch velocity of 3409 m/s. Therefore (ignoring decimals), exactly the same velocity as if you launched horizontally.

From both of the above ways of looking at it, then, a vertical launch is tempting because, in theory at least, you need (almost) exactly the same delta-v to get to Minmus as you need - in reality - to get into LKO to start with.

The problem, however, is TWR and gravity losses. There is simply no way you can launch straight up at 3409 m/s - you will burn up instantly in the atmosphere.
You can't even launch straight up with a TWR of 10: you're accelerating upwards at 90m/s² and you're going to burn up after about 20 seconds (going 1900m/s at 20km altitude) unless you have a serious heat protection.
However, plugging the edge of the atmosphere back into that "classic" equation gives a velocity of 3223 m/s on leaving the atmosphere. From the equation distance=1/2(acceleration x t²) and velocity=acceleration x t, we can get to that needed velocity at 70km with an upwards acceleration of about 74m/s² which corresponds to a TWR of 8.55. It will take 43 seconds to reach space and gravity losses over that time will be 426 m/s. You'll still need serious heat protection, but it should be manageable.
In other words, with a very high TWR it can be tempting to look at the numbers and think that going straight up makes sense.

Of course, there are two major reasons that this approach fails, and fails utterly.
The first reason is drag. A TWR of 8.55 will have you going at around 1000m/s at 8km and 2000m/s at 25km. Drag losses will be huge - hundreds of m/s.
The second reason is cost. It is far more expensive and heavy to build a ship with such a high TWR, and protect it against heating.

 

So is some sort of compromise possible? With a TWR of 6, you reach a "Minmus-capable" velocity (3150m/s) at 101km. Gravity losses are 630 m/s and you go a bit slower through the atmosphere (1000 m/s at 10km).
And finally, for Minmus, the point where you break even by going vertically is with a TWR of 4. With that, you reach a "Minmus-capable" velocity (3030m/s) at 156km. It takes you 102s to get there and so gravity losses are 1010 m/s. And you only break the 1000m/s barrier at about 17km altitude.

Therefore, with a TWR less than 4, you'll always be better off with a non-vertical approach.

And remember that this is only for very distant destinations. Going vertically, your orbital velocity on arrival is zero - a difference of under 50m/s compared to an efficient orbital approach for Minmus, but the difference approaches 200m/s on arrival at the Mun, and that difference will be felt in the capture burn.

[Pecan]

I hate launching anything with an M700.

Anyway ... for me, all launches circularise in Kerbin orbit just to get their breath back.  It's a good photo opportunity for the newer recruits too, while they wait for their tug or ferry to collect them.
Yes, it is slightly cheaper to burn directly to your destination instead of stopping in Kerbin orbit before transferring.  The latter is an awful lot easier and more reliable though and both of them are hugely expensive compared to building a reusable infrastructure.

[Snark]

8 hours ago, Spricigo said:

That is the whole point. No matter what you do gravity will be there doing its thing. Instead of trying to overcome gravity the idea it's to flight in a such way that the effect of gravity will not hinder you. Once you have enough orbital speed you will not be falling toward kerbin instead you will be falling around kerbin. 

It's kind like not swing against the stream pushing you away from the beach but getting out the stream in the first place.

Well, yes.  There are good theoretical reasons why a gravity turn works better than a vertical ascent.  And these arguments have been made before, and clearly the OP is aware of them, and clearly was unconvinced by them:

12 hours ago, bewing said:

Yes, there are a bunch of oldtimers who are going to moan and say "not this issue again!!?" Tough. Each new group of newbies needs to be exposed to the argument...

So, I'm in the direct ascent camp. The gravity turn zealots make wild assumptions about how wonderfully efficient it is to do it their way. I say the gains are minimal at best

^ Which demonstrates the issue here:  When you have an assertion that's based on scientific theory-- no matter how sound the theory-- then not everyone is going to be convinced.  Folks who disagree can dismiss the arguments as "wild assumptions".

This kind of argument pops up from time to time on the forums-- witness a recent lengthy discussion about reaction wheel placement.  (TL;DR:  Turns out that the physics people are correct, after all.)

To be clear:  I'm not trying to dismiss or belittle the I-don't-buy-your-arm-waving-theory, I-need-a-practical-demonstration people.  The "messy" parts of KSP (e.g. atmospheric launch) are as much engineering as physics, and it's not always obvious which physical factor is going to win, when you mix in the flexibility of rocket design and complex phenomena such as aerodynamic drag.  And also, not all people think and solve problems alike:  there are plenty of people for whom seeing-is-believing (and, more to the point, not-seeing is not-believing), who work better from practical examples more than theory.  There's nothing wrong with that-- I kinda have a foot in both camps myself, being both an engineer and a physics major.

And even the most sound theoretical argument benefits from a practical demonstration.

Accordingly, kudos to @bewing for framing the question in practical terms, this was a great idea. 

 

Clearly the OP's expectation turned out to be not just wrong, but dramatically wrong:  gravity turn was not just better than vertical ascent, but a lot better (by a factor of more than three).  Which is great!  I think one of the great joys of KSP is its ability to teach us new and unexpected things.  For me, there are few sweeter things in life than the "eureka" moment that happens when a cherished but incorrect assumption is broken.

 

 

5 hours ago, Pecan said:

I hate launching anything with an M700.

...and amen to that, brother!  That thing is such a pain.  (Which is part of its charm, IMHO.  Gives a nice engineering challenge to design around it.)

[Red Iron Crown]

I like this thread just because of the "let's not talk theory, show me" approach to demonstrating efficiency.  

It would be more interesting still if only a payload was specified rather than a complete upper that can be used during the mission. Something like one or two full 1.25m ore tanks to a specified orbit around the Mun.

It would also be interesting to try it on a vacuum world, where atmospheric heating isn't an issue so the direct ascent approach can dial up the TWR for lower gravity losses. In theory direct ascent doesn't need as energetic a transfer orbit (Pe can remain below the surface), it might be interesting to see if that factor is outweighed by the additional engine mass. I suspect it is but would be interesting to see it demonstrated.

[Snark]

1 hour ago, Red Iron Crown said:

It would also be interesting to try it on a vacuum world, where atmospheric heating isn't an issue so the direct ascent approach can dial up the TWR for lower gravity losses. In theory direct ascent doesn't need as energetic a transfer orbit (Pe can remain below the surface), it might be interesting to see if that factor is outweighed by the additional engine mass. I suspect it is but would be interesting to see it demonstrated.

On a vacuum world, a vertical ascent can dial up the TWR to minimize gravity losses.  But on the other hand, a gravity turn can crank it over to horizontal much sooner, since it doesn't have to worry about climbing steeply out of that pesky atmosphere.

For most vacuum worlds, it's pretty close to a moot point, since gravity is so low that even a very modestly designed craft can have a (local) TWR of 10 or more, and going beyond that you get diminishing returns.  They're easy enough to escape that it probably doesn't matter a whole bunch whether you go up or sideways.

The place where it would really make a difference would be a heavy vacuum world, where both the gravity and the escape velocity are significantly big.  Tylo's about the only really good example in the game for that, I think, though I suppose Moho ought to be enough to tell the difference.

[OhioBob]

1 hour ago, Red Iron Crown said:

It would be more interesting still if only a payload was specified rather than a complete upper that can be used during the mission. Something like one or two full 1.25m ore tanks to a specified orbit around the Mun.

I agree, the payload should just be some inert mass.  It should be up to the rocket designer to determine what propellant, engines, and stage configuration he wants to use.

By the way, I was curious to see if I could beat bewing's cost per kilogram using a liquid fuelled first stage rather than cheap SRBs.  I was successful, but not by a lot.

8444 launch cost + 340 LF cost = 8784 total cost / 6208 kg to Mun orbit = 1.415 funds/kg

One of the things that ran my cost up was that I included four AV-R8 Winglets at a cost of 2560.  With some practice I might be able to eliminate those, or replace them with something cheaper, to lower my cost further.

(edit)

Made some changes and got it down to:

6184 launch cost + 353 LF cost = 6537 total cost / 6129 kg to Mun orbit = 1.067 funds/kg

 

Edited August 16, 2016 by OhioBob

[Spricigo]

7 hours ago, Snark said:

9 hours ago, Red Iron Crown said:

It would also be interesting to try it on a vacuum world, where atmospheric heating isn't an issue so the direct ascent approach can dial up the TWR for lower gravity losses. In theory direct ascent doesn't need as energetic a transfer orbit (Pe can remain below the surface), it might be interesting to see if that factor is outweighed by the additional engine mass. I suspect it is but would be interesting to see it demonstrated.

On a vacuum world, a vertical ascent can dial up the TWR to minimize gravity losses.  But on the other hand, a gravity turn can crank it over to horizontal much sooner, since it doesn't have to worry about climbing steeply out of that pesky atmosphere.

For most vacuum worlds, it's pretty close to a moot point, since gravity is so low that even a very modestly designed craft can have a (local) TWR of 10 or more, and going beyond that you get diminishing returns.  They're easy enough to escape that it probably doesn't matter a whole bunch whether you go up or sideways.

The place where it would really make a difference would be a heavy vacuum world, where both the gravity and the escape velocity are significantly big.  Tylo's about the only really good example in the game for that, I think, though I suppose Moho ought to be enough to tell the difference.

That wold be a interesting demonstration indeed. 

Just some consideration about the relevance of the demonstration  from gameplay perspective ( I see few plus and a few minus) :

+ higher TWR is open (and useful) to both approaches. With the possibility of more similar designs we may have a better comparison of the flight profiles. 

+allows for very aggressive gravity turn,  thus being also a opportunity to compare the importance of thrust Vs ISP for gravity turn. 

-in the regular game mission starts at kerbin.  In practical gameplay how we get to Tylo/Moho/Whatever it's much more decisive factor to how we get out of there. 

-Once in space there is opportunity to refuel and/or docking with a tug,  at this point reusability and recoverability become as important (if not more)  than individual launch cost. 

- Reliability is almost never mentioned in those frequent discussion about efficiency.  If the situation where one method is better than other don't happen often enough the question has little practical value. 

-How much the efficiency 'cost' in funds,  time and effort is also often let out of these discussions.  And there I think @bewing have a point. A particular design and method will be ineffective before the results overcome the cost,  yet quite often we bash methods/design ideas to people struggling to understand it. 

[Norcalplanner]

Wow, this thread is squarely in my wheelhouse. Sorry I didn't see it before now.

Check out the tutorial link in my signature for lots of tips and tricks for Cheap and Cheerful rocket design (such as omitting fins due to their cost). When I get to a real computer, I'll post some links to other threads, including my recent cost-per-ton of payload to orbit challenge, and some tests where I examined various TWRs and ascent profiles.

[GoSlash27]

#1 judging your efficiency purely by DV numbers is misleading. That says nothing about what kind of launcher you will have to build, how much fuel it will expend, or what it will cost.

#2 burning to LKO and then transferring to an intercept is mathematically more efficient, even when judging purely by DV numbers.

Best,
-Slashy

[Norcalplanner]

As promised, here are some helpful links which discuss this issue:

SRB discussion from my tutorial.  This was done back in 1.0.4, so it's a bit out of date.  Based on some more recent testing, I wouldn't recommend thrust limiting SRBs as much.  I now shoot for an initial TWR between 1.5 and 1.9.

The Cheap and Cheerful Rocket Payload Challenge.  This was back in 1.0.5, but it showed just how cheaply payloads could be lofted to orbit, in most cases by using clusters of Kickbacks.  Many entries were under 700 funds/ton.  A select few were under 600 funds/ton. Here's a good launch video showing a 617 funds/ton design by maccallo.

A Degree Makes a Difference - or How I Learned to Stop Worrying About Drag and Concentrate on Gravity Losses. This is a recent 1.1.3 thread where I look at the effects of different TWRs and ascent profiles in RSS.  I ended up coming around to the "crank it over as hard as you can so long as you still make orbit and don't burn up in the atmosphere" school of thought.  Gravity losses are at least an order of magnitude greater than aerodynamic drag losses.  FAR changed things a bit, but not as much as I thought it would.

[foamyesque]

5 hours ago, GoSlash27 said:

#2 burning to LKO and then transferring to an intercept is mathematically more efficient, even when judging purely by DV numbers.

Well, more efficient than going straight up, yes. But the most efficient way is to use your ascent to put you directly into a Hohmann with a minimal periapsis (30km-40mk) to your desired intercept, skipping the LKO part. That manouver requires slightly less deltaV, since you don't bother to raise the periapsis, and because you do your burn lower down you get more benefit from Oberth.

[Superfluous J]

22 minutes ago, foamyesque said:

Well, more efficient than going straight up, yes. But the most efficient way is to use your ascent to put you directly into a Hohmann with a minimal periapsis (30km-40mk) to your desired intercept, skipping the LKO part. That manouver requires slightly less deltaV, since you don't bother to raise the periapsis, and because you do your burn lower down you get more benefit from Oberth.

Mathematically yes, but practically speaking I personally save more dV by setting up a maneuver node in LKO, over going right to Mun from launch and then having to fix my imperfect trajectory midway there.

It's the difference between theory and practice. In theory, there's no difference. In practice, though...

[Diche Bach]

ADDIT: Oh wait! Gravity TURNS you're asking about. I thought this thread was about gravity-ASSISTS (as in using Mun close flyby to get to Minmus cheaper).

There are players who question the efficiency of gravity turns? Really, I just assumed that everyone used those. You literally just go STRAIGHT up to orbit!?

Excuse my simplistic biologists conceptions of gravity, but why one Kerbin would you imagine that is in anyway "competitive" with doing a gravity turn? You're fighting Kerbin's gravity 100% when you are thrusting normal and I'd guess that at 45% pitch relative to the horizon you are fighting it about half as much, somewhere closer to 0% when at a pitch close to horizontal, no?

Once I learned how to basically play the game, I try to generally fly by intuition, and forego all the maths and efficiency analyses, though I can certainly understand how that is appealing.

My general procedure on a craft without special considerations or on a mission without special considerations:

1. Launch at about 1.3 to 1.5 TWR

2. Lower TWR to the 1.2 to 1.35 range during the first few seconds (I frequently have a single throttle controlled engine that has enough oomph to handle the whole thing which is in the same stage as a few SRBs which are also powerful enough to lift the whole thing, although perhaps marginally at the initial launch, so launch may be: start with throttle at 1/3, taper back to nearly zero [to retain the gimbal from the liquid fueled booster] then as the SRBs deplete, throttle back up to hit the 1.2 to 1.3 ballpark.

3. Try to keep acceleration under control until about 35,000 m (thick air = resistance so more power = heat and not > velocity). 300 m/s at 35,000 seems about right, and from there, acceleration can proceed at will without much impact from air resistance (all gut intuition here, no background research done to confirm/deny this)

4. I try to be about a 60-degree pitch by the 25 to 35,000 ball park, 50/45 ballpark pitch by the 45,000 ball park, and generally just leave it there until apo gets past 70,000m maybe pitching down a bit more toward the horizon, but I've been burned a few times where my orbital insertion burn was too shallow and I wound up losing my upward momentum and had to whip the ship back around to normal to regain upward velocity a few times so I avoid flattening out the trajectory too soon.

5. Once I'm past 70,000m, and my apo is perhaps 85,000 or 90,000 (at least) and velocity sufficient that apo is increasing "rapidly" going flat to push out peri.

Now contrast this with setting up the first couple "short-range only" RemoteTech satellites, where the initial ascent was much more steep (retained nearly vertical pitch until about 50,000 or 55,000m, and only a minimal turn from there till 70,000 when fragile gear is deployed and the "horizontal" burn to push out the peri is undertaken . . . these launches certainly seemed to use up more fuel overall.

Based on these general principles "working" I doubt I'm patently wrong on any of this, though might well be confused or "less than optimal" on most or all of it. However, one thing that remains outside my "intuition grasp" is the issue of TWR once one is past the pesky atmosphere.

Obviously some times you want your burn to happen FAST, because you want to take advantage of the effects of doing a burn at a very specific point in your orbit, and in those instances large dV changes are better suited by a short, sharp high TWR burn.

But what about: I'm in space, I really don't need to move my apo out any farther, I just want to get my peri to ~71km as "cheaply" (meaning as much fuel as possible left over): what is optimum then?

TWR in the 1.05 ballpark? or as much as one can muster?

heading pointed slightly above horizon? right at horizon, or perhaps even slightly below?

I have some payloads that are really pushing the limits of what is doable for a given launchers size and TWR (think top heavy . . . I'll have to post files of my space station core and modules 2, 3, and 4, which are all in orbit now, but which were quite "delicate" to get up there). The gravity turn (and its benefits in terms of fuel use) have to be moderated by the need to: a. get past the thicker parts of the atmosphere, but not too fast that resistance and turblence are created too much; b. wait to turn until velocity is high enough that forward momentum "overrides" the radial momentum created by the turn. On these, I generally "quadruple" the initial thresholds of my "standard" launch = start gravity turn at 250 m/s instead of 50 m/s; only shoot for ~80degree pitch at 25,000 to 35,0000 (instead of 60 degree). Even in the upper atmosphere and in LKO, a really top heavy ship has to be handled differently or it can get itself into trouble.

Wow, I was just wondering this exact same question, and came on here to post a thread that addressed these very same questions!

Edited August 18, 2016 by Diche Bach
I misunderstood the nature of the thread

[Red Iron Crown]

2 hours ago, Diche Bach said:

3. Try to keep acceleration under control until about 35,000 m (thick air = resistance so more power = heat and not > velocity). 300 m/s at 35,000 seems about right, and from there, acceleration can proceed at will without much impact from air resistance (all gut intuition here, no background research done to confirm/deny this)

Your gut is leading you astray. A not-stupidly-draggy craft will get to orbit most efficiently by going as fast as possible as low as possible without burning up. Gravity losses trump drag losses in nearly every case, so getting sideways as quickly as possible is key to high efficency. (All this from a delta-V-to-orbit perspective.) Early stages should almost never be throttled back.

Also: I think you may be conflating "normal" with "radial" in your post.

[foamyesque]

It is possible to have cases where your drag losses are an important consideration, but they're not common. I ran into that issue when I was building super tiny orbiters, which had noticeable drag losses at 60km.

[Norcalplanner]

I should have some new runs on stock Kerbin done later today or tomorrow with various TWRs and ascent profiles.

[Diche Bach]

Ah thanks Red Crown! You know, after jotting all that out, I've been having a look at leaving it full throttle and I had already started to observe the error of my ways. Thanks for confirming!

Full throttle from here on!

Also, yes, RADIAL is where you are pointing when you are looking "straight up" isn't it

[numerobis]

@Diche Bach: to a first approximation, gravity loss when you're firing 45 degrees is only 30%.

At 45 degrees you're getting about 70% of your burn going into horizontal speed, and 70% vertical. If you assume all your vertical is lost, then you burned 100% and got 70%, so your loss is 30%.

Generally, the relation is 1-sin(pitch). The loss is 50% at 60 degrees, 30% at 45 degrees, and a mere 14% at 30 degrees pitch.

The assumption is wrong when you don't let your vertical speed go to zero (i.e. you don't go into LKO), as @bewing proposes. So my formula is just an upper bound on gravity loss.

[Norcalplanner]

New comparative ascent data for stock Kerbin is now up:

 

[Stoney3K]

On 18/08/2016 at 11:08 PM, Red Iron Crown said:

Your gut is leading you astray. A not-stupidly-draggy craft will get to orbit most efficiently by going as fast as possible as low as possible without burning up. Gravity losses trump drag losses in nearly every case, so getting sideways as quickly as possible is key to high efficency. (All this from a delta-V-to-orbit perspective.) Early stages should almost never be throttled back.

Also: I think you may be conflating "normal" with "radial" in your post.

Define "stupidly draggy". I usually can't veer off more than 2-3 degrees from my prograde angle before my craft start flipping over, and that's with a simple rocket design which has a payload in a fairing, and 3 or 4 strap-on boosters.

When your TWR is massive it means your angle of attack is going to increase as well if you want to make a turn at low altitude. Usually when I can't turn fast enough due to high TWR, I will burn the lower stages, coast and finish the turn, and then ignite the upper stage until Ap.