When to start Gravity Turns

[Colonel_Panic]

I usually start at 8k for most rockets. Some faster rockets I go a bit higher to kill the drag faster so I don't waste as much fuel on drag, and slower rockets I sometimes start my turn earlier, since they're not moving fast enough for drag to be a major issue, and I want to get my lateral movement up.

The real trick is completing your turn at the right point. You can save a considerable amount of fuel if you surf the upper atmosphere gaining speed until your periapsis pops out from under you. A perfect burn is one you don't have to circularize until the other side of the planet, and only takes a few m/s of delta v when you do.

[Tristavius]

As with most, nearer 10k for lighter craft, nearer 18 for heavier seems to be the sweet spot... it also depends a bit on the staging, for example my two everyday lifters (rated for about 40 and 70 tons) both shed several tanks/engines at around the 15k mark, if I turn before this they have a habit of clipping something as they come off (there's invariably a degree or two of wobble after the turn) so I wait until that part of the staging is over before turning.

I've also changed my methodology recently; I was taught (for better or worse) to do a small gravity turn, push my apoapsis to 80k then stop; as I approach the apoapsis turn the rest of the way and begin to circularize my orbit. Recently I've discovered following my prograde marker after the gravity turn and never really cutting the engines unless I really over push my apoapsis seems to be a lot more reliable when getting designs with borderline delta v into orbit.

[Nuke]

An answer right in the middle of the reasonably optimal range for all but the most unusual designs (8 to 12km) is certainly not "absolutely wrong" Mr. "Real KSP Pro."

But if you want to be absolutely wrong, keep calling it a gravity turn. What you and pretty much all KSP players do is in no way a gravity turn, it's simply a pitch maneuver (assuming you even bothered to orient the craft in a standard flight attitude where pitch controls your climb angle).

A real gravity turn starts as soon as you leave the launch pad.

i believe the proper astronaut lingo is "roll program".

[Nibb31]

I start my gravity turn very gently at 1000m and end it at 70000m.

There is no reason it should start at 8 or 15km...

[drakesdoom]

The really short answer is where it works for that vehicle.

The long answer is given a body with no atmosphere.

The true answer would have to include the above equation, atmospheric drag, thrust limitations of the vehicle, staging of the vehicle, and structural limitations of the vehicle. Good luck.

Also the Manley answer

If you don't want to watch he says for most rockets start at 10km and don't take too long to turn over.

[Scialytic]

The long answer is given a body with no atmosphere.

The true answer would have to include the above equation, atmospheric drag, thrust limitations of the vehicle, staging of the vehicle, and structural limitations of the vehicle. Good luck.

If you're considering that much effort, you might as well just test it empirically in the game. With an approximate gradient-descent approach, you could figure out the optimum altitude to start pitching over in relatively few launches (depending on how precise an answer you want).

[Felsmak]

I typically start at 10 000 meters; 20 000 if the drag is too harsh on my current rocket.

[RoboRay]

i believe the proper astronaut lingo is "roll program".

The roll program is different. The shuttle was not positioned on the launchpad in a way that that they could simply pitch the nose slightly down toward the horizon to start their gravity turn. They had to roll the craft first, shortly after clearing the tower, to get the craft in the proper "wings aligned with the horizon and the vehicle inverted (crew head-down)" orientation for the turn.

Edited May 17, 2013 by RoboRay

[RoboRay]

I start my gravity turn very gently at 1000m and end it at 70000m.

There is no reason it should start at 8 or 15km...

It depends on the curvature of the turn. I start mine at 1km and make a very gradual turn, still at 80 degree climb angle when I clear 10km, so it's effectively vertical from a drag-reduction standpoint.

However, most KSP players seem to prefer just flopping their craft right over to 45 degrees instantly (giving up efficiency to steering losses in the process). For them, waiting to 10km (+/-2km) is essential. It's not as good a way to do it, but it's easier.

Edited May 17, 2013 by RoboRay

[rosenkranz]

I typical start at 5km but i don't just dump it over to 45 deg. I inch it over so that i'm at 45 deg at about 15km up so as to minimize steering losses and balance drag losses.

[Rune]

Despair, you mortal, if you try to pull a correct gravity turn manually in KSP. No, seriously, it is pretty much impossible unless you let a computer do it. I depends heavily on the T/W curve of the vehicle in question, and it involves heavy calculus to come up with the trajectory graph. Mechjeb gives a good enough approximation, form what I've heard, and I'm pretty sure I keep gravity losses reasonably low by hand. The only general advice is to try to thrust always close to your prograde vector, the rest depends on the rocket.

Rune. T/W is usually glossed over, and it's mighty important.

[rkman]

Absolutely wrong. Scott Manley said it varies per Launch Vehicle but is mostly at 10km. There's more science than that. Sure, 10km is standard, but I'm looking for the science and formulas behind it.

Sounds like 10k is about right rather than absolutely wrong.

[DerekL1963]

The roll program is different. The shuttle was not positioned on the launchpad in a way that that they could simply pitch the nose slightly down toward the horizon to start their gravity turn. They had to roll the craft first, shortly after clearing the tower, to get the craft in the proper "wings aligned with the horizon and the vehicle inverted (crew head-down)" orientation for the turn.

Yes, and no. The Shuttle rolled not only to control aerodynamic loading (and to point the antennas on the Orbiter's upper surface at the downrange ground stations*), but also to align the stack with the intended flight azimuth. The Saturn V rolled for the latter reason as well. (If you read the Apollo Flight Journals (http://history.nasa.gov/afj/) you'll see the calls for the roll program somewhere around T+00:00:15 to T+00:00:30.)

The reasons why they needed to align the vehicle with the launch azimuth are complicated and buried deep in the arcana of guidance system design and are related to the low performance of guidance electronics back in the early days. (Watch the Polaris and Poseidon launches starting at about 00:10 in this video (http://youtu.be/uljVI4m5e3c), and you'll see them perform a roll program almost immediately upon ignition.) Nowadays it's an anachronism kept out of habit, if they do it at all.

*Later in the program, some of those downrange ground stations were decommissioned for budgetary reasons. If you compare the time of the roll back to 'heads up', you'll see where the later launches had to roll back earlier to point those antenna at the TDRSS birds.

Despair, you mortal, if you try to pull a correct gravity turn manually in KSP. No, seriously, it is pretty much impossible unless you let a computer do it. I depends heavily on the T/W curve of the vehicle in question, and it involves heavy calculus to come up with the trajectory graph. Mechjeb gives a good enough approximation, form what I've heard, and I'm pretty sure I keep gravity losses reasonably low by hand. The only general advice is to try to thrust always close to your prograde vector, the rest depends on the rocket.

This. There is no One True Way to do a gravity turn.

Edited May 17, 2013 by DerekL1963

[RoboRay]

I was trying to describe the physical orientation of the craft, not listing reasons for doing the maneuver.

As the real KSC isn't on the equator like the Kerbal version (nor is Vandenberg, Wallops, Baikonur, Jiuquan, etc...), the minimum orbital inclination is the launch-sites latitude. The inclination can be increased (if desired) by adjusting the heading north or south (depending on what the target plane is), which is part of what the roll maneuver is doing (orienting the craft so that only a pitch maneuver is needed to start the turn, rather than a combination of pitch and yaw).

It's definitely not just done out of habit.

Edited May 17, 2013 by RoboRay

[Colonel_Panic]

pro mode: keep straight up until a little ways into the ascent, then just pitch your nose over slightly and peg your prograde vector for the rest of the ascent.

If you don't make it to orbit, you started too soon. If you have to correct by burning radially near the end of your ascent, or have a long circularizing burn, you started too late. With a little trial and error, you'll discover the perfect time to do it... for that particular rocket.

[DerekL1963]

I was trying to describe the physical orientation of the craft, not listing reasons for doing the maneuver.

The name "roll manuver" is rooted in the historical reasons for performing the maneuver, and this leads to confusion among some as to what exactly the vehicle is doing during the maneuver and why it's doing it. (Doubly so since the name of the maneuver has remained the same, even though the definition has changed somewhat.) There's two different things going on simultaneously, for two different reasons. The first is rolling the vehicle (for reasons I go into below), the second is pitching/and or yawing the vehicle onto the desired azimuth. (For reasons which should be obvious.)

As the real KSC isn't on the equator like the Kerbal version (nor is Vandeberg, Wallops, Baikonur, Jiuquan, etc...), the minimum orbital inclination is the launch-sites latitude. The inclination can be increased (if desired) by adjusting the heading north or south (depending on what the target plane is), which is part of what the roll maneuver is doing (orienting the craft so that only a pitch maneuver is needed to start the turn, rather than a combination of pitch and yaw).

You can reach *any* inclination from *any launch site* without rolling - only pitch and yaw are required. There's no mathematical reason to roll so as align the vehicle so the gravity turn is done only about the pitch axis. Dig up the equations - there isn't even any term for roll in them.

That's where the historical reasons for the name come into play. Back in the old days pitching and yawing onto the desired azimuth could degrade guidance system accuracy (at best) or cause the guidance system to hiccup (at worst) if the vehicle pitch plane was not coplanar with the guidance pitch plane.* The electronics of the day simply weren't good enough to quickly and accurately perform the required coordinate transformation if the vehicle's pitch plane was not coplanar with the guidance stable platform pitch plane, so they maneuvered the vehicle as soon as possible to bring them coplanar. This makes the gravity turn to take place completely in pitch, and minimizes any errors that might be caused by inaccurate or lagging** coordinate transformation. So the term "roll maneuver" came to be applied, regardless of when the pitch over (the gravity turn) in the orbital plane actually begins. (And it begins at different altitudes for different vehicles, because initial T/W differs as does T/W over time as different vehicles consume different fuels and oxidizers (I.E. LH2 doesn't weigh the same as RP-1) at different rates.)

Modern guidance systems however are accurate enough that they can easily deal with the body and guidance axes not being coplanar, but the name has stuck regardless of whether they actually roll or not. The Trident-I and -II for example, do not roll. (They don't even *have* a roll control system active until the bus separates from the third stage.) But as you can see in the video I linked above, the Polaris and Poseidon do roll - why? Because their guidance systems weren't as accurate and couldn't handle the coordinate transformation.

It's definitely not just done out of habit.

Rolling (other than for Shuttle, which needed to roll in order to control aerodynamic loads and for antenna positioning.) is done out of habit, because space programs are in general are conservative as hell. Pitching and/or yawing onto the desired inclination is done because otherwise "you aren't going to (your desired orbit in) space today".

* Even today, for accuracy reasons, the guidance stable platform is oriented to the final orbital plane, but the vehicle's orientation is tied to the physical orientation of the launch pad. That's why the Complex 39 launch pads are oriented the way they are, so that the tower clearance maneuver started pitching them over onto their final orbital heading. Their performance margins were so slim that this was thought necessary to minimize the amount of time the F-1 engines were gimbaled off vertical (from the vehicle body axis) as that reduces performance.

** Yes, lag is an issue in real space programs too... that's what caused the 1202 alarm during Eagle's final approach - the computer was overloaded and starting to lag because they had an extra (and unneeded at that phase of flight) program running and eating memory and CPU cycles.

[austin4050]

Does not have to be at 10km but around that area. If it is quite heavy do it around 13-19Km.

[RoboRay]

You can reach *any* inclination from *any launch site*

You can not directly launch into an orbital inclination lower than your launch-site latitude. It's physically impossible. What you can do is launch to a minimum-inclination trajectory and make a plane change later, at a lower latitude, to attain a lower inclination. This can even be done as part of your launch, if you desire.

If you desire an inclination equal to or greater than the launch-site latitude, you can launch directly into it without a post-liftoff plane change, which is why I made the distinction.

without rolling - only pitch and yaw are required. There's no mathematical reason to roll so as align the vehicle so the gravity turn is done only about the pitch axis.

Yes, this is in agreement with what I said. I didn't say there was a mathematical reason it was done, just that it's done to simplify the guidance solution to maneuvering around a single axis.

And doing things conservatively because you know that method works reliably doesn't make it merely "out of habit." If it was purely out of habit, it would be a matter of whimsy, not conservativeness with incredibly expensive equipment (and sometimes lives at stake).

Edited May 17, 2013 by RoboRay

[AngelLestat]

yeah, many times with heavy rockets i do it at 20 or 24km some times.

All depends on the speed and rocket design.

If you use some nuclear engines you want to gain height fast, so maybe 30 k is not out of the question.

Also if you had too much drag, even if your rocket is fast, is convenient do it late.

I guess everybody does it a little earlier what they should becoz they take the shuttle example.

But the shuttle needs that inclination early on to balance the tank weight.

[NeoMorph]

One thing peeps here have missed... What if it is on the Mun... Gravity turn there (and all no or low atmo density planets and moons) you can start your gravity turn right after takeoff.

So gravity turns depend on your vehicle AND your location.

[Joe_Bender]

pro mode: keep straight up until a little ways into the ascent, then just pitch your nose over slightly and peg your prograde vector for the rest of the ascent.

If you don't make it to orbit, you started too soon. If you have to correct by burning radially near the end of your ascent, or have a long circularizing burn, you started too late. With a little trial and error, you'll discover the perfect time to do it... for that particular rocket.

There it is!!!!

[Space_Coyote]

Personally I turn at 10% of the final atltitude.. so here's an example..

if I say go for 150,000 meters (150 km) then my gravity turn would be at 15,000 meters or 15km..

I find that usually when I get into space using that formula, I tend to get enough reserve fuel that I can actually be able to finish the burn without wasting fuel oon the upper stages..

Just a point of thought for folks to try..

Space_Coyote

[JiWint]

I start it at second atmosphere layer (darker one),it's around 11km

[JiWint]

Also,it depends on a planet/moon you are taking off.

If it has atmosphere,you have to do it as late as you can,(not too late actually),because when you turn horizontally,there's even more drag.

If it doesn't have an atmosphere,you can take off almost horizontally!

[thescientist]

It's impossible to give a "right height" because it change a lot: depends on mods (as far), celestial body, TWR, weight...

Just to give you an idea: my minimalistc design to minmus give the best if you start at 5km, my ultra-heavy payload lifter must start at 15 km to lift up to 200 tons (using FAR)