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

[Red Iron Crown]

1 minute ago, Stoney3K said:

Define "stupidly draggy".

In this context it means "with no attempt to streamline at all, with many draggy parts exposed to the airstream". It takes a very draggy craft to make drag losses outweigh gravity losses (at least in the stock KSP system, in RSS with its higher speeds the balance is a bit different).

1 minute ago, Stoney3K said:

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.

These are design issues rather than theoretical ones, but even so it is better to get turned sooner. When I use a high TWR craft with insufficient control authority to start the gravity turn in the way I want, I just angle the craft on a launch clamp to compensate (example). 

[Diche Bach]

There is a difference between draggy and being radially asymmetric ("top heavy") [although some drag queens no doubt demonstrate that the two are NOT mutually exclusive . . . pic foregone for PG rating].

The issue Stoney3K is referring to (one I encounter frequently myself) is I think not an issue of drag, it is an issue of imbalance between the radial and anti-radial mass in the spacecraft.

If your spacecraft has excessive radial mass ("excessive" meaning exceeding some threshold of its attitude control facilities to prevent it from suffering multiplied yaw or pitch attitude changes) relative to its anti-radial mass at any point in its flight, this means that all the fancy calculations about ideal attitude per TWR can get tossed right out the window.

Like any experiment, the intent is to focus on the effect of a small number of independent variables (TWR, pitch sequencing) and issues like radia asymmetry, structural integrity, and even drag are assumed to be constant, i.e., the same craft was used for all tests with different values of the independent variables.

Of course, the ideal is to NOT build radially asymmetrical vessels, but constraints in terms of funds, or available techs may well make that ideal difficult to achieve. For example: I've just archived my old career play in order to try for better performance with a smaller mod set. I'm just in my first couple missions and experimenting with various cheap, low-tech designs to achieve some of the early contracts: Test TK-38K radial decoupler; Test LV engine; leave atmosphere; orbit Kerbin, etc.

I'm constrained by a number of factors: VAB not upgraded = cannot launch a proper orbiter (too many parts); no reaction wheels = control constraints that limit how effectively cheap boosters like hammers can be used; etc. Here the issue of radial asymmetry comes in to play and the findings of the "Gravity turn significantly moderates dV to orbit" hypothesis testing are merely instructive not formulaic, because my ability to build craft that take advantage of the findings are limited.

[Diche Bach]

So here is a question: why does Kerbal Engineer considering turning a craft which is pointing straight up (radial) toward the east using the "yaw to starboard" key (D) to be "pitch" and not "yaw?" Is this technically correct?

Technically, shouldn't "pitch" be the angle of the spacecraft relative to normal/anti-normal?

And why does it consider the "belly" to be indicative of the heading at least until after one has pitched/yawed a bit away from a perfectly radial angle?

Obviously it doesn't baffle me so much I am unable to cope, but those discrepancies just don't make sense to me.

[Stoney3K]

When you're going straight up, pitch and yaw are pretty ambiguous. Since there is no way to tell in which direction the 'up' axis of the spacecraft is travelling horizontally, you can't determine a heading. In this case, pitch and yaw are just based on assumptions on your spacecraft design, which by default, is pointing its 'roof' due north and its belly due south (the pitch axis). Usually people do a 90 degree roll before making a pitch down maneuver to start their gravity turn.

Only when you have some horizontal velocity (not travelling vertical anymore) pitch and yaw actually start to have their respective meaning and there is a sense of heading.

[OhioBob]

In KSP, the pitch/yaw axes are defined by the orientation of the probe core/command pod.  For instance, the hatch of a Mk1 signifies the "pitch up" direction.  This makes sense because in real life a spacecraft has defined pitch and yaw directions determined by how the inertial guidance and attitude control systems are set up.

A real life rocket's alignment on the launch pad is typically govern by the orientation of the launch tower and the location of access panels and connection points (such as fueling lines, power cables, etc.)  Therefore the alignment of a rocket's pitch/yaw axes as it sits on the launch pad is rarely correct for its planned flight azimuth.  This is why a rocket performs a roll before it starts to pitch down.  It rolls until its pitch axis is perpendicular to the flight azimuth, then it simply has to pitch down to fly along its intended heading.

In KSP the default launch pad alignment is has the pitch axis pointed east-west and the yaw axis pointed north-south.  If you plan an eastward launch and you want to literally "pitch down" during ascent, then you need to roll 90 degrees before commencing your pitch down maneuver (or rotate the vehicle in the VAB before sending it to the launch pad).  However, I think most people dispense with the roll and perform the maneuver by "yawing right".  Although we commonly refer to this as a pitch down maneuver*, we are technically rotating in yaw (D key = yaw right).

I think the reason KSP does this is obvious - so that the W,A,S,D keys are aligned to the movements of the rocket that we see on our computer screens.  If we roll the rocket 90 degrees, then either the keys are misaligned to the observed movements, or we have to rotate the image so we are looking eastward.  And if we rotate the image, then we have a difficult time judging the degree of tilt of the rocket (must rely on the NavBall).

(edit)
*  We are pitching down in relation to the horizon even though we may be yawing in terms of the rocket's rotational axes.

Edited August 25, 2016 by OhioBob

[OhioBob]

2 hours ago, Stoney3K said:

which by default, is pointing its 'roof' due north and its belly due south (the pitch axis).

Actually the pitch axis is pointed side-to-side (east-west) through the cockpit.  If we pitch north or pitch south, then we are rotating around an axis that is aligned east-west.  The pitch axis is normal to the pitch plane.

(edit)  Also, the default launch pad alignment has the roof south and the floor north.
 

Quote

Usually people do a 90 degree roll before making a pitch down maneuver to start their gravity turn.

Really?  I've always figured there's only a small number of purists who bother performing a 90 degree roll (those who take the "pitch down" thing literally).  I suspect that the vast majority of KSP players just yaw right during launch.

Edited August 24, 2016 by OhioBob

[foamyesque]

Depends on what I'm driving, but for the most part, yeah, I'll either just yaw or rotate the rocket 90 degrees in the VAB.

[numerobis]

Pitch can be defined relative to the vessel, or relative to the horizon. Both are valid definitions.

The pitch-over is about the horizon, because thrust and gravity don't care which way you define up to be; gravity itself defines up.

Controls tend to be based on the vessel, because otherwise, which way is "down" when you're pointing straight up? (Also it's more natural for anyone who's played a flight sim -- or flown or driven or captained.)

[OhioBob]

1 hour ago, numerobis said:

Pitch can be defined relative to the vessel, or relative to the horizon. Both are valid definitions.

The pitch-over is about the horizon, because thrust and gravity don't care which way you define up to be; gravity itself defines up.

Controls tend to be based on the vessel, because otherwise, which way is "down" when you're pointing straight up? (Also it's more natural for anyone who's played a flight sim -- or flown or driven or captained.)

That's true.  But pitch-over is also about control of the vessel.  That's the reason for the roll maneuver - to bring the rocket's pitch/yaw axes into the correct orientation relative to the flight azimuth and horizon.  Therefore, when we want to pitch the nose down toward the horizon, we're rotating the rocket about only one of its control axes - usually pitch, though it could be yaw.  If the axes weren't correctly aligned, then pitching the hose down toward the horizon would require rotating the rocket around two of its control axes.

[Stoney3K]

I never understood why the default orientation for the launch pad is south-north which is perpendicular to the ship's direction as it would roll out of the VAB. A lot of players launch into equatorial orbits for anything beyond Kerbin's SOI, there are only a few scenarios where a S-N launch orientation would make sense (like polar orbits).

The interesting thing is, the Space Shuttle launches in the same direction and rolls to face 'belly up' so it's actually upside down. Is that because of the aerodynamic properties of the shuttle and the external tank?

[LN400]

46 minutes ago, Stoney3K said:

The interesting thing is, the Space Shuttle launches in the same direction and rolls to face 'belly up' so it's actually upside down. Is that because of the aerodynamic properties of the shuttle and the external tank?

I have noticed but never really thought about why. Could it be to use the CoM/thrust offset for a less fuel hungry g-turn? With the shuttle hanging underneath the tank, the drag and mass will allow for less thrust from the shuttle's engines than if it rode on top, kind of thing?

[numerobis]

7 hours ago, OhioBob said:

If the axes weren't correctly aligned, then pitching the hose down toward the horizon would require rotating the rocket around two of its control axes.

Fun trick: if, on reentry, your control wheels aren't quite managing to hold your pitch (relative to horizon) steady, roll 45 degrees. Then SAS can use both axes, so it's 40% stronger. You also use double the electricity, so be careful not to run out.

Unfortunately the autopilots that I see (MechJeb, kOS, SAS) treat the two axes as completely separate controls. That prevents using spin stabilization to deal with asymmetries.

[Rodhern]

I too think it is an interesting question @bewing asks. Just how much will a gravity turn save you, when a near vertical ascent will do the job too?

I know @bewing doesn't care about the price in delta-v (fuel units used), so I will leave my old delta-v budget comments and budget questions for another thread.

As @Stoney3K and @Diche Bach already hinted at, often it is way easier to design and fly a craft capable of near vertical ascent compared to one that will fly a more ideal trajectory to orbit.

For instance, when launching (low-tech) probes with Remote Tech, you might not have a connection to the probe during ascent. You can still do your launch as a simple timed sequence, and I don't think the (slight?) extra fuel cost should put you off.

Edited August 25, 2016 by Rodhern
fixed the "@"-quotes

[OhioBob]

5 hours ago, Stoney3K said:

I never understood why the default orientation for the launch pad is south-north which is perpendicular to the ship's direction as it would roll out of the VAB.

Actually the orientation in the VAB and the orientation on the launch pad is the same.  It's the camera angle that changes.  In the VAB the default camera angle is looking east, while on the launch pad the default camera angle is looking north.  This gives the impression that the vehicle rotates 90 degrees when you go from VAB to launch pad, but it really doesn't.

As for why they use the orientation that they do, it's so that the rocket moves in directions that correspond to the layout of the command keys on the keyboard.  If you want to tilt the rocket to the right (east) then you press the rightmost command key (D).  It's the most intuitive way to do it.

 

[Diche Bach]

So technically, if you execute your gravity turn by yawing to starboard, it _is_ yaw, but it is called "pitch" to keep it more confusing/consistent 

I'll have to create a Kerbal Engineer template that includes all the pitch/roll/yaw output that are there for aircraft and use those when I launch a rocket and see what it looks like.

I also enjoy how the space ship in orbit will whip around 180-degrees at some particular point in its orbit (only for certain types of high-inclination orbits I think) while on 50x warp speed  First time I saw that I was all stressed out . . . then I realized it was just buzzing right along fine and did it again next lap!

Using stuff like this, someone could do a pretty hilarious montage video with Yakkity Sax as the theme song.

[Red Iron Crown]

2 hours ago, Diche Bach said:

I also enjoy how the space ship in orbit will whip around 180-degrees at some particular point in its orbit (only for certain types of high-inclination orbits I think) while on 50x warp speed

That's the camera whipping around, not the ship. The camera goes wonky near the poles in some modes.

[Tyko]

10 hours ago, Stoney3K said:

The interesting thing is, the Space Shuttle launches in the same direction and rolls to face 'belly up' so it's actually upside down. Is that because of the aerodynamic properties of the shuttle and the external tank?

9 hours ago, LN400 said:

I have noticed but never really thought about why. Could it be to use the CoM/thrust offset for a less fuel hungry g-turn? With the shuttle hanging underneath the tank, the drag and mass will allow for less thrust from the shuttle's engines than if it rode on top, kind of thing?

 

From NASA.gov

Jerry from London (UK): Why does the shuttle rotate through its axis immediately after take off and how is it achieved?

Leinbach: Well, we need to rotate the shuttle through its axis to get on the proper azimuth for the launch inclination itself. If you look at the launch pad out by the Atlantic Ocean, it's in a due north/south orientation. So if we didn't roll right after lift-off, the orbiter would be going due east. We don't like that. We need to align the orbiter with the International Space Station inclination. So, after lift off we have to roll to achieve that heading as the orbiter continues through ascent. It's achieved by gimbling the main engines itself – the three SSME – the main engines – and also by gimbling the two solid rocket booster nozzles to achieve that roll profile.

Edited August 25, 2016 by tjt

[Corona688]

It's almost the same question as "how much longer are the two straight lines than the one angled line?"

It's plain that the angled line is the shortest distance from "lower right" to "upper left".  The angled line is more or less what you follow on the gravity turn, direct ascent follows the straight lines.

It's even worse than that, however, as the lines aren't actually straight but curved.  Direct ascent ignores the curves and stays on course by brute force, gravity turn follows the curve.

Edited August 25, 2016 by Corona688

[Stoney3K]

17 hours ago, tjt said:

 

From NASA.gov

Jerry from London (UK): Why does the shuttle rotate through its axis immediately after take off and how is it achieved?

Leinbach: Well, we need to rotate the shuttle through its axis to get on the proper azimuth for the launch inclination itself. If you look at the launch pad out by the Atlantic Ocean, it's in a due north/south orientation. So if we didn't roll right after lift-off, the orbiter would be going due east. We don't like that. We need to align the orbiter with the International Space Station inclination. So, after lift off we have to roll to achieve that heading as the orbiter continues through ascent. It's achieved by gimbling the main engines itself – the three SSME – the main engines – and also by gimbling the two solid rocket booster nozzles to achieve that roll profile.

That was not what I was really wondering about. Why is the orbiter ascent upside down instead of flying SRBs down, cockpit up orientation?

(Also, time to scale up the Kickbacks and ModuleManager them into having proper gimbals, so we can make decent shuttles.)

@Corona688: Your chart shows why a gravity turn is more efficient for LKO. But the more the trajectory is 'vertical' as opposed to circling around Kerbin (ie. going to Mun, Minmus or interplanetary) the horizontal line becomes more and more tiny and therefore less relevant.

Edited August 26, 2016 by Stoney3K

[Corona688]

5 hours ago, Stoney3K said:

 

@Corona688: Your chart shows why a gravity turn is more efficient for LKO. But the more the trajectory is 'vertical' as opposed to circling around Kerbin (ie. going to Mun, Minmus or interplanetary) the horizontal line becomes more and more tiny and therefore less relevant.

Not really.  Just because you're in Kerbin orbit, doesn't mean you've actually escaped gravity.  I'll have to launch a gravioli to check, but the force of gravity in near orbit is should be nearly as large as the force on the ground, as it is on Earth;  it's only the constant fall which creates "free fall" conditions.  Meaning, escape is still quite a long journey away.

You can do it, of course.  With a good enough TWR, the time losses are minimized (which is how I managed to accidentally escape not just Kerbin's but Kerbol's orbit via two BACC's and a probodyne.)  And the aerodynamics improvement which made gravity-turn so efficient also makes direct ascent more efficient.  But going sideways means you get to use a really good, light, efficient engine, not the highest thrust monster you have.

So, direct ascent will be much worse for some vehicles than others.  If you've got a terrific TWR you can get away with it.  A creaky monster which claws up at 1.2 gees all the way would feel the losses of a direct ascent badly.

Edited August 26, 2016 by Corona688

[Stoney3K]

For an interplanetary transfer or a transfer to the edge of Kerbin's SOI, it doesn't really matter as much anymore. Why bother with estabilishing a parking orbit if you're just going to leave it anyway to escape Kerbin's gravity well altogeher? When you're burning for another planet, you might as well burn straight up at the proper moment.

[foamyesque]

2 hours ago, Stoney3K said:

For an interplanetary transfer or a transfer to the edge of Kerbin's SOI, it doesn't really matter as much anymore. Why bother with estabilishing a parking orbit if you're just going to leave it anyway to escape Kerbin's gravity well altogeher? When you're burning for another planet, you might as well burn straight up at the proper moment.

 

That's incorrect, because you still save on deltaV and needed TWR by burning horizontally in a direct ascent mode than straight up. This is the exact argument @bewing made in his OP, you realize.

[Eric S]

2 hours ago, Stoney3K said:

For an interplanetary transfer or a transfer to the edge of Kerbin's SOI, it doesn't really matter as much anymore. Why bother with estabilishing a parking orbit if you're just going to leave it anyway to escape Kerbin's gravity well altogeher? When you're burning for another planet, you might as well burn straight up at the proper moment.

 

Pythagoras still matters, just not the way you think.  We can all see that the specific orbital energy between two orbits with a Mun-level apoapsis, with one a center-of-Kerbin periapsis and the other a low-Kerbin-orbit periapsis, favors the former but is minimal.  The point of the non-vertical ascent is that it's adding to its orbital energy in a more efficient manner.  It should also be pointed out that the extra energy of the latter orbit isn't entirely wasted as it means that you're closer to matching the Mun's orbital velocity when you get there.

2g worth of vertical thrust yields 1g worth of acceleration (two steps forward, one step back).  2g worth of prograde thrust 45 degrees off of the horizon yields 1.73g worth of acceleration (1.43 steps forward, one step back, and 1.43 steps to the side).  Now, as you increase the TWR, this effect diminishes, but it doesn't go away until you're talking an infinite TWR.  Even at a TWR of 10.0, the vertical ascent is still losing about 10% more acceleration to gravity. 

This doesn't mean that the entire first case launch will be 73% more efficient because even a gravity turn spends some time going close enough to vertical that it doesn't really benefit from this during that time.  In addition, as the craft burns fuel, it's TWR goes up, so even a low-TWR craft should have a higher TWR by the time it has pitched over enough for this to make a significant difference.  On the other hand, as the craft starts reaching orbital velocity, the craft will wind up losing even less acceleration.  But really, given that the specific orbital energy between the two target orbits only differs by about 3%, it doesn't take much to overcome that.

Yes, you can say that you'll just design a craft with a higher TWR, but then you're using a more expensive craft to launch the mission, which is a better method of determining efficiency than pure delta-v.  

But all of this is the theoretical math that got dismissed by the OP, so really, we need to fix the test case and rerun the tests.  As others have mentioned, we need a defined payload to be delivered but treated as inert or at least mostly inert.  If there's a probe core there I don't mind the idea of it controlling the craft, but if it has fuel and/or an engine, neither should be used or altered.

[Rodhern]

1 hour ago, Eric S said:

... 2g worth of prograde thrust 45 degrees off of the horizon yields 1.73g worth of acceleration (1.43 steps forward, one step back, and 1.43 steps to the side).

Hi Eric, how do you arrive at 1.73g ?

[Superfluous J]

Yeah I get 1.474. Well, calculator.net does

http://www.calculator.net/triangle-calculator.html?vc=45&vx=2&vy=1&va=&vz=&vb=&angleunits=d&x=66&y=16

("Down" in the diagram is actually toward the upper right, and the rocket is thrusting right to left)