Gravity turn. Do you need to do it?

[Scrogdog]

Ah, but only if the goal is difficult.

Right now, yes. But at some point, in career mode, economics will matter.

Someone made the point the other day that, theoretically, launch windows never close. It all depends on available energy. Practically, launch windows DO close. They more or less close on our current technological levels.

I suppose that just because something is possible does not make it wise.

[LegoDino77]

In gravity turn, does it matter how fast you go under 10 km?

[Pzar]

In gravity turn, does it matter how fast you go under 10 km?

You don't want to break terminal velocity

There's a table here: http://wiki.kerbalspaceprogram.com/wiki/Kerbin

As for the burning straight up vs gravity turn, the real savings in the gravity turn method comes from changes in design you can make.

Burning straight up you need a TWR > 1 until you reach escape velocity.

Once you're in orbit, you need a TWR > 0 to achieve escape velocity.

So while the delta V requirements are not significantly different, the last ~1k or so (based on PirateAE's numbers, and ~4.5k to orbit, and you could probably save over a little more than that if you trie) can be done with engines with a much higher ISP, meaning less fuel spent.

[Krizzen]

I got it! This has been a mystery to me, but I think I have it. So, a fuel optimal ascent is one that doesn't waste fuel to atmospheric drag and one that doesn't fight gravity longer than it should. Two very simple concepts. As examples, you could have an apoapsis of 500km and *still* fall right back to Kerbin. Also, you could burn so much fuel in the atmosphere fighting drag that you *never achieve orbit*. The ad-hoc gravity turn is a (somewhat) optimal solution to give you an apoapsis far enough outside the atmosphere that you won't waste all your fuel to drag inside the atmosphere. It's like a compromise.

Now, this is a rather basic way to look at it, though I haven't been able to conceptualize it without all the insightful posts.

I'm sure there is a mathematical solution (I'd love to know what it is!).

I think the most significant gains are from your ship's trajectory. Again, you go straight up, and you will fall back down even with a very high apoapsis, just like throwing a rock straight up as high as you can. If you aim too low, you will waste all your fuel fighting to burst out of the atmosphere, like an airplane. If you aim just right, your trajectory will pierce the thickest part of the atmosphere letting you accelerate where the atmosphere is the thinnest and at the point where you are "falling" back to the surface.

[haltux]

The ad-hoc gravity turn is a (somewhat) optimal solution to give you an apoapsis far enough outside the atmosphere that you won't waste all your fuel to drag inside the atmosphere. It's like a compromise.

No, sorry to insist but no. I just don't want people to get the wrong ideas after this thread. What you say is perfectly true when the goal is to orbit. But here the goal is to reach escape velocity. There is no evidence that orbiting helps in terms of fuel consumption to reach escape velocity. It somehow makes sense to do that because you need less thrust on upper stage and so you can have smaller engines. But that is a side question. At the simple, purely physical question: give a 1 stage rocket, what is the best way to escape with the least possible fuel consumption, the answer is not "orbiting as low as possible". The answer is, I believe, "go vertical in the atmosphere and then prograde", which is almost a vertical ascent on Kerbin.

I think the best approach now would be to go to some actual science forum and find some guy who would do the math for us.

Again, you go straight up, and you will fall back down even with a very high apoapsis

Well at some point you reach escape velocity, and you do never fall back. That's the whole point of the thread.

Edited June 3, 2013 by haltux

[Darren9]

I think there's a bit of misuse of the term "Gravity Turn" going on. To achieve an orbit you need to add both horizontal and vertical velocity from the start position at rest on the planet surface, since you start pointing up you need to turn on the way up to add horizontal velocity. That's what we're doing in KSP, turning the rocket on the way up attempting to find the most efficient path to the desired orbit, and there's no escaping it, you have to add a horizontal component or you can't achieve an orbit.

Normally a craft wont just turn itself, you need to spend energy rotating it (RCS, activate a control surface which increases drag, gimbal engines and divert some thrust to rotate the craft, use the magical rotation power of command pod/SAS units). A gravity turn achieves the necessary direction change to make orbit without any energy expended by the craft to rotate it (apart from the initial turn to set it up), you make one control input to initiate the turn and then no other input. The effect of the CoM not being directly above the CoT turns the craft "for free" with no input/energy use (gravity pulls the nose down). It's a complex calculation though and what makes rocket science rocket science . If your not making one initial control input to set the calculated angle for that particular craft and then no other directional input (no RCS, gimbal or SAS, only adjusting throttle and staging) to arrive at your desired orbital height perfectly horizontal then you aren't performing a "Gravity Turn", you're just making the necessary turn to achieve orbit and spending some energy to do it. It's a small saving, hardly noticeable in KSP (with a perfect gravity turn you wouldn't need RCS at the top, the craft would just fall into pro-grade for you), but in real space travel small savings are still massive amounts of money.

The result seems to be though that nobody is actually calculating and gaining a benefit from a true "Gravity Turn", we're just making the necessary direction adjustments to make orbit by expending energy (in KSP you get free rotation energy in some cases) to turn the craft and calling it a "Gravity Turn" and claiming we're gaining from it with "you need a gravity turn". Really it's a trajectory optimisation that's beyond the scope of the game. You only need to turn if you want to achieve an orbit, and there's many reasons in real life to want an orbit first as already mentioned.

The problem with testing in KSP is different craft perform differently in the same ascent profiles. I did some testing as well. In an attempt to negate all other variables I used a single stage with constant ISP (the same in atmosphere and vacuum) and MechJeb'd it with "Limit to terminal Velocity". Under those conditions straight up appears to be the winner, it seems to be expected (to me at least) - straight up has the lowest loss to drag (shortest course through atmosphere) and there's no gain in staging/ISP changes that orbit first may offer. As soon as you make a design with staging/varying ISP you can have a craft that will perform better going to orbit first. This is the problem I think, you can design two different craft and each will perform better/worse going either way so there's no correct general answer as to which you should use. Without staging/isp, drag/atmospheric differences with specific craft designs both ways should be equal??? (you can'ny break the laws of physics Jim!)???

[Fox21]

getting by all the confusing math. Doing a gravity turn around 10,000meters saves feul

[asae]

I see there's something that most of the posts here forgot to mention, the oberth effect that can be exploited by gravity turn to orbit first, the lower in the gravity well you are the less delta V you will need to change your velocity.

[cesarcurado]

Before I just did straight to orbit. I found out about gravity turn while messing around with mechjeb for the first time (I don't like to use MJ and don't anymore).

The difference is quite noticeable as your trajectory is already kinda wide the time you get to orbit so it saves you some fuel.

[Darren9]

I see there's something that most of the posts here forgot to mention, the oberth effect that can be exploited by gravity turn to orbit first, the lower in the gravity well you are the less delta V you will need to change your velocity.

I'm struggling with this,

In astronautics, the Oberth effect is where the use of a rocket engine when travelling at high speed generates much more useful energy than one at low speed.

You shouldn't go at high speed inside the atmosphere, above terminal velocity wastes fuel? Is there something I'm missing?

[asae]

Sorry I didn't quite word that well, what I meant to say was that using a gravity turn to get into a low orbit allows you to then use the oberth effect to your benefit but it can only be done when in orbit, not inside the atmosphere.

[Sir Nahme]

I'm struggling with this,

You shouldn't go at high speed inside the atmosphere, above terminal velocity wastes fuel? Is there something I'm missing?

the closer you are to the planet (your periapsis) the less thrust it takes to increase your velocity

the difference is noticable even with a difference of 10km (you accelerate faster at a periapsis of 70km over one of 80km)

to achieve the necessary delta V to escape an orbit altogether it then behooves you for efficiencies sake to make a single orbit.

You accelerate towards your periapsis and decelerate towards your apopiasis

it's easier to increase your speed while you are already accelerating than it is to do so while you decelerate

this is because in one instance (heading towards periapsis) you are using gravity to gain speed already and thus aren't fighting it, but when heading away you are fighting it

[Darren9]

the closer you are to the planet (your periapsis) the less thrust it takes to increase your velocity

the difference is noticable even with a difference of 10km (you accelerate faster at a periapsis of 70km over one of 80km)

to achieve the necessary delta V to escape an orbit altogether it then behooves you for efficiencies sake to make a single orbit.

You accelerate towards your periapsis and decelerate towards your apopiasis

it's easier to increase your speed while you are already accelerating than it is to do so while you decelerate

this is because in one instance (heading towards periapsis) you are using gravity to gain speed already and thus aren't fighting it, but when heading away you are fighting it

But when you're gaining speed you're loosing height (gravitational potential energy?) and when you're loosing speed you're gaining height. Combined kinetic and gravitational potential is still equal at any point on the orbit? And for maximum efficiency you should act at either Ap or Pe when acceleration/deceleration is zero? Maths hurts my head

[Sir Nahme]

Losing height isn't a problem if you aren't intersecting the atmosphere and losing energy

Thats the wonder of an orbit, you aren't losing energy as long as you keep missing the planet. The potential energy stays the same. You can simply add to it easier as approaching periapsis

[tavert]

In an attempt to negate all other variables I used a single stage with constant ISP (the same in atmosphere and vacuum) and MechJeb'd it with "Limit to terminal Velocity". Under those conditions straight up appears to be the winner, it seems to be expected (to me at least) - straight up has the lowest loss to drag (shortest course through atmosphere) and there's no gain in staging/ISP changes that orbit first may offer. As soon as you make a design with staging/varying ISP you can have a craft that will perform better going to orbit first. This is the problem I think, you can design two different craft and each will perform better/worse going either way so there's no correct general answer as to which you should use. Without staging/isp, drag/atmospheric differences with specific craft designs both ways should be equal??? (you can'ny break the laws of physics Jim!)???

Controlled testing is the right way to answer this if you won't believe the intuition-based explanations people are giving. Straight up is the lowest loss to drag, but the HIGHEST loss to gravity. Gravity losses will cost GM/r^2 - v_horizontal^2/r. The second term there is centripetal acceleration, in a circular orbit the two terms cancel out and you're not fighting gravity anymore.

Another issue with going straight up is that you have to spend energy to bring the mass of all your unused fuel up to high altitudes. If you make a "gravity turn" (either perfect AoA = 0 like in the real world or our Kerbal approximations), get into low orbit then burn to escape, you never had to bring any of that propellant higher than 70 km, so you get more kinetic energy (escape velocity!) and didn't have to pay a cost in potential energy.

Here's my controlled test. 1-man lander can, one X200-8 tank, one X200-16 tank, one aerospike. Craft file if you're lazy: https://dl.dropboxusercontent.com/u/8244638/Escape%20Test.craft

Pictures here http://imgur.com/a/f5Syg

It has 5610 m/s vacuum delta-V, 5581 atmospheric. It starts with a low enough TWR that it never hits terminal velocity, so don't have to worry about that here. It takes about 4600 m/s for this design to get to a 71 km orbit, leaving 1000 m/s which is enough to escape Kerbin from low orbit. But if you expend all of its 5600 m/s delta-V going straight up (first picture), you won't escape Kerbin at all, your apoapsis will top out at 3.5 million km.

Are we done here?

Edited June 3, 2013 by tavert

[Darren9]

Controlled testing is the right way to answer this if you won't believe the intuition-based explanations people are giving. Straight up is the lowest loss to drag, but the HIGHEST loss to gravity. Gravity losses will cost GM/r^2 - v_horizontal^2/r. The second term there is centripetal acceleration, in a circular orbit the two terms cancel out and you're not fighting gravity anymore.

Another issue with going straight up is that you have to spend energy to bring the mass of all your unused fuel up to high altitudes. If you make a "gravity turn" (either perfect AoA = 0 like in the real world or our Kerbal approximations), get into low orbit then burn to escape, you never had to bring any of that propellant higher than 70 km, so you get more kinetic energy (escape velocity!) and didn't have to pay a cost in potential energy.

Here's my controlled test. 1-man lander can, one X200-8 tank, one X200-16 tank, one aerospike. Craft file if you're lazy: https://dl.dropboxusercontent.com/u/8244638/Escape%20Test.craft

Pictures here http://imgur.com/a/f5Syg

It has 5610 m/s vacuum delta-V, 5581 atmospheric. It starts with a low enough TWR that it never hits terminal velocity, so don't have to worry about that here. It takes about 4600 m/s for this design to get to a 71 km orbit, leaving 1000 m/s which is enough to escape Kerbin from low orbit. But if you expend all of its 5600 m/s delta-V going straight up (first picture), you won't escape Kerbin at all, your apoapsis will top out at 3.5 million km.

Are we done here?

Thank you for explaining, I didn't consider at all the cost of carrying fuel higher. My ill-conceived test craft actually doesn't, it was two nacelles strapped to an X2-1600 for 9500 Delta-v and 28 TWR. The vertical escape burn is over at very low altitude, below 70 Km in fact, probably with your explanation and hindsight it was a particularly inappropriate test vehicle. It does still seem to be able to escape with higher delta-v remaining in a vertical ascent. I understand now that that's bending a rule with unrealistically high thrust engines rather than proving something though.

[tavert]

The higher your TWR, the longer you'll be able to keep up with terminal velocity so the more difference drag will make. Also MechJeb's "limit to terminal velocity" only considers your vertical speed, so it doesn't always make sense any more after you start turning. I bet you were seeing re-entry flames when you did a turn with that high of a TWR. To reduce drag losses you'd probably need to keep the throttle down with that thing until you're up to at least 20-30 km, maybe even higher.

And gravity losses only account for 1/28'th of the thrust of your example, vs over 87% initially with mine. An SSTO example has its TWR change pretty dramatically over the course of a flight, but you could for example make an asparagus launcher that maintains a more consistent TWR and see how much of a difference it makes there.

I also don't know what you mean by nacelles, but I'm going to guess it's some overpowered mod engine.

Edited June 4, 2013 by tavert

[Darren9]

The higher your TWR, the longer you'll be able to keep up with terminal velocity so the more difference drag will make. Also MechJeb's "limit to terminal velocity" only considers your vertical speed, so it doesn't always make sense any more after you start turning. I bet you were seeing re-entry flames when you did a turn with that high of a TWR. To reduce drag losses you'd probably need to keep the throttle down with that thing until you're up to at least 20-30 km, maybe even higher.

And gravity losses only account for 1/28'th of the thrust of your example, vs over 87% initially with mine. An SSTO example has its TWR change pretty dramatically over the course of a flight, but you could for example make an asparagus launcher that maintains a more consistent TWR and see how much of a difference it makes there.

I also don't know what you mean by nacelles, but I'm going to guess it's some overpowered mod engine.

The nacelles are indeed OP mod engines. It seems to be moving around 1000m/s at 20Km up for the orbital one, a bit faster or slower if I increase or decrease the turn shape. It coasts up to Ap from around there as well. The vertical one is around the same to that point, then it accelerates faster and it's all done around 50Km and coasts all the way out. I've tried some different ascent paths for the orbital one and I can't seem to make it beat the vertical. Is it mathematically impossible for a vertical ascent to beat an orbit first regardless of TWR and it's just me (or Mechjeb) making an inefficient ascent to orbit or have I just randomly stumbled upon a legitimate exception? When you mentioned the cost of carrying fuel higher, for this particular craft the vertical only uses all the fuel it needs by 50Km, the orbit first carries the circularize and escape fuel up to 70Km. I know it's OP and couldn't really exist, I'm just curious about why I'm seeing what I am.

[Sma]

All that matters for me is the fact that pretty much all real life launches do some form of gravity turn. For all I know there is some gravity mechanic that makes it nearly impossible to do a straight up launch in r/l, unless you fight the rotation of the planet(once you leave the atmosphere the ground will be rotating below you). Which I think where the gravity turn comes in handy.

There is a reason "gravity turns" generally head in an easterly direction as opposed to a westerly direction. You get a boost from the planets rotation by doing a gravity turn. You may have to fight more air friction but this is why you don't do the turn as soon as you launch, you do it somewhere between 10km and 20km(depending on your launch craft and your launch profile/plan). When I do gravity turns I don't go straight up over to 45 degrees right away, I try to do it in steps, say 10 degrees every 1km - 2km after passing 10-15k (or the transition between the light blue to the dark blue on the right of the atmosphere gauge.

As for shooting straight up and reaching escape velocity the trick is you have to maintain that escape velocity to escape the SOI. Other was if you turn off your engines when you reach escape velocity and you're going straight up gravity will start pulling you back down at which point you're not traveling at escape velocity any more. Then I imagine if you do manage to leave the SOI of say Kerbin, you end up in orbit of the sun, likely a very long/narrow elliptical orbit assuming you don't fall into the sun that is.

Of course, all that being said, I'm not an expert and am not claiming to be one. All that babble is just how I see it.

[tavert]

The nacelles are indeed OP mod engines. It seems to be moving around 1000m/s at 20Km up for the orbital one, a bit faster or slower if I increase or decrease the turn shape. It coasts up to Ap from around there as well. The vertical one is around the same to that point, then it accelerates faster and it's all done around 50Km and coasts all the way out. I've tried some different ascent paths for the orbital one and I can't seem to make it beat the vertical. Is it mathematically impossible for a vertical ascent to beat an orbit first regardless of TWR and it's just me (or Mechjeb) making an inefficient ascent to orbit or have I just randomly stumbled upon a legitimate exception? When you mentioned the cost of carrying fuel higher, for this particular craft the vertical only uses all the fuel it needs by 50Km, the orbit first carries the circularize and escape fuel up to 70Km. I know it's OP and couldn't really exist, I'm just curious about why I'm seeing what I am.

As I said, the higher your TWR, the less gravity matters and the more drag matters. Are you using MechJeb 1 or MechJeb 2? If MechJeb 1, try turning off the auto-throttle as soon as it starts turning, and manually use a much lower throttle. In MechJeb 2, try limiting the throttle % or the acceleration m/s^2 once it starts turning. You should be able to use less fuel to reach escape velocity that way, I believe it should beat the straight up ascent. But maybe your TWR is so high that it doesn't work the way I think it should, I'd have to do a numerical trajectory optimization to be sure.

[mrfox]

If you look at this graphically...

... going straight up just doesn't make sense.

[haltux]

There is a reason "gravity turns" generally head in an easterly direction as opposed to a westerly direction. You get a boost from the planets rotation by doing a gravity turn.

Sure. I think "vertical asent" should be unerstood as "prograde ascent", that is going prograde once you leave the atmosphere. Practically, prograde ascent is very close to vertical both on earth and kerbin.

As for shooting straight up and reaching escape velocity the trick is you have to maintain that escape velocity to escape the SOI.

No you don't. If you reach the escape velocity (which is 11 km/s at the surface of earth and decreases with altitude), you never come back to earth even if you switch off engines. You escape earth forever, assuming you are not orbiting another body which brings you back to earth at some point. That's the definition of escape velocity.

[Sma]

No you don't. If you reach the escape velocity (which is 11 km/s at the surface of earth and decreases with altitude), you never come back to earth even if you switch off engines. You escape earth forever, assuming you are not orbiting another body which brings you back to earth at some point. That's the definition of escape velocity.

That's true, guess I wasn't thinking very clearly (it was late, or at least that is my excuse . Though you still orbit something, when you get a Kerbin escape you'll be orbiting the sun. I suppose you could then also burn to escape the suns gravity as well. However I did a test going straight up and cut my engines after escape velocity(3,431.03 m/s). It said I was leaving Kerbin before I reached escape velocity, but I kept going until I was faster than 3,431.03 m/s. After cutting engines I started slowing down again, though I think the ship would have left Kerbins SOI before it slowed down below EV.

So I guess theoretically if you don't have a specific target in mind going straight up works just fine. The problem I'd imagine comes when you want to go to the Mun, Minmus or another planet. I think it'd take a lot more planning to get the timing just right for the launch.

As for the slingshot idea, if it didn't work real space agencies wouldn't be using it to get their craft into deep space. I can understand the thinking "but you have to spend energy to get to a higher orbit". This may help in understanding it better. http://en.wikipedia.org/wiki/Oberth_effect I guess it works in that little fuel is needed to extend your AP higher. Then once you're coming "down" from your AP, with your engines off (saving fuel) you keep gaining speed faster and faster. You can try it out, put a ship in, lets say a 100km x 500km(or 1000km) orbit and see what your speed is at PE with engines off, time warp until you get close to PE again, stop time warping and see what the speed is at PE this time. More than likely it'll be faster, although maybe not a lot faster, it just depends on how far out your AP is. Obviously at this point if you start firing your engines again you'll end up with a faster velocity than you could have with the same amount of fuel, or at least that's the idea.

With all that said (however correct/incorrect) What it really comes down to is what works best for you. I'll stick to my gravity turns to reach orbit then go from there.

[maltesh]

As Haltux implies, Escape velocity is dependent on your current distance from the center of the object you're escaping; it's not a single value that applies everywhere within the gravity well.

Assuming a free, unpowered trajectory, if you're at a velocity higher than escape velocity for your distance at any point on your trajectory, you'll be at above escape velocity for the distance at every other point on your trajectory.

[Rage097]

A gravity turn is almost always necessary. Unless you want a very fuel inefficient rocket, i'd suggest gravity turning would be an excellent idea.