The advantages of launching straight up

[farmerben]

In KSP it's possible to launch straight up until entering the next SOI, or leaving Kerbin SOI.  This saves fuel (maybe 25-35%) versus achieving low orbit and then leaving the SOI.

In reality we could build an Artemis mission that shoots straight up to the moon, with no changes in direction.  Ideally this would work with something like the falcon heavy with more boosters.   It's easy to land the boosters at the launch site no matter how high they go.  The central stage can fly all the way to the moon where the central booster will be put into a graveyard, either in orbit or on the surface.  The same will even work for going to Mars, align the rocket to vertical at sunrise and just go up.  Although this plan is not fully reusable the central core stages will be valuable salvage for future colonies.

[SunlitZelkova]

30 minutes ago, farmerben said:

The same will even work for going to Mars, align the rocket to vertical at sunrise and just go up.

China sort of did this with Tianwen-1. The Long March 5 rocket did not put the spacecraft into Earth orbit, it launched on a direct trajectory to Mars.

This is just a guess, but I think China probably did this due to the trauma experienced with Fobos-Grunt (which also carried Yinghuo-1, China’s first attempted Mars orbiter) getting stranded in its parking orbit rather than any technical necessity.

Now this is just some simple (dumb?) reasoning based on KSP, but launching directly would involve getting to orbital velocity + trans-*destination* injection velocity anyways, so why not go to orbit?

I’m sure there are other issues I am missing, or we would already be doing this.

[sevenperforce]

Yes, we could design a rocket that performs a direct burn to the Earth-Moon Lagrange Point 1 (L-1) without first achieving low earth orbit (LEO).

It would be wildly, wildly more Δv-expensive than performing a pitchover and entering orbit, whether that orbit was a parking orbit with a subsequent trans-lunar injection (TLI) burn or a so-called "direct ascent" using a continuous burn to avoid any coast period.

The Δv required to go from Earth's surface to LEO varies based on ascent profile, but is typically considered to be between 9.3 and 10.0 km/s. The Δv from LEO to C3=0 (that's Earth escape) is 3.22 km/s, while the Δv from LEO to trans-lunar injection is 3.20 km/s. The escape velocity from Earth's surface is 11.86 km/s, so I will assume for the purpose of this calculation that the Δv required for a direct burn from the Earth's surface to L-1 is approximately 11.84 km/s.

Now if you look at the above numbers, then you'd think that the "just burn straight up" approach seems better, because a direct burn is 11.84 km/s while going to LEO first will cost you between 12.5-13.2 km/s. However, you're forgetting that the direct burn to L-1 is not instantaneous; it takes time. If your launch vehicle can accelerate at somewhere between 1.5 gees and 3.0 gees, then it's going to take between 400 and 800 seconds of time to do that. Let's say you have a LOT of thrust on your launch vehicle and so you can do all of this in just 500 seconds. Well, that means you're going to be pushing against Earth's gravity for 500 seconds, and Earth's gravity is going to pull you back down at 9.81 m/s². Accordingly, your required Δv is going to go up by 4.9 km/s, bringing total dV to 16.74 km/s.

So you're WAY better off going to orbit first.

The reason for this, as many here probably already know, is that gravity drag drops off as you ascend to an orbit, thanks to centripetal acceleration.

7 minutes ago, SunlitZelkova said:

China sort of did this with Tianwen-1. The Long March 5 rocket did not put the spacecraft into Earth orbit, it launched on a direct trajectory to Mars.

This is just a guess, but I think China probably did this due to the trauma experienced with Fobos-Grunt (which also carried Yinghuo-1, China’s first attempted Mars orbiter) getting stranded in its parking orbit rather than any technical necessity.

I’m sure there are other issues I am missing, or we would already be doing this.

What China did with Tianwen-1 is the same thing we did with Pioneer-4 when we launched it on the Juno II: a direct ascent into low earth orbit that continues directly into the injection burn without SECO or a coast period.

No launch beyond Earth has ever gone straight up without at least momentarily entering low earth orbit.

[Superfluous J]

51 minutes ago, farmerben said:

This saves fuel (maybe 25-35%) versus achieving low orbit and then leaving the SOI.

This is not strictly true.

You can make a rocket for which it's true, but for each of those you can make a cheaper one that goes to orbit first.

[kerbiloid]

13 hours ago, farmerben said:

In KSP it's possible to launch straight up until entering the next SOI, or leaving Kerbin SOI.  This saves fuel (maybe 25-35%) versus achieving low orbit and then leaving the SOI.

This Direct Ascent trajectory was being planned for lunar missions since 1950s, but

1. It requires more fuel.

2. It doesn't forgive errors. You can't slide around the Moon and return like Apollo-13 did. If you crash, then you crash.

13 hours ago, farmerben said:

In reality we could build an Artemis mission that shoots straight up to the moon, with no changes in direction. 

SLS barely can deliver something in the economic trajectory mode, let alone the direct ascent.

Edited October 27, 2022 by kerbiloid

[RCgothic]

To add on to what @sevenperforcesaid, straight up is also a less efficient thrust direction.

A rocket that could accelerate at 1.5G straight up could accelerate at ~2.3G horizontally whilst maintaining the vertical thrust necessary to counter gravity.

This is the main reduction in gravity drag - getting pitched over as soon as possible increases acceleration, which decreases time under thrust. The shorter the time fighting gravity, the less energy is wasted.

Reduction in apparent gravity due to centripetal acceleration also allows smaller, less massive engines to be used on the upper stage. Reduction in dry mass has an exponential effect on the final achievable speed.

Burning straight up is the single most expensive way to get out of a gravity well.

[KSK]

Going to LEO first also gives you time to make doubly sure that the spacecraft is in good working order after the climb to orbit before heading out to the moon. Maybe not the issue it was during the Apollo days but still helpful I think, at least for crewed flights.

Edited October 27, 2022 by KSK

[sevenperforce]

9 hours ago, kerbiloid said:

This Direct Ascent trajectory was being planned for lunar missions since 1950s, but

1. It requires more fuel.

2. It doesn't forgive errors. You can't slide around the Moon and return like Apollo-13 did. If you crash, then you crash.

So there are three different things here, actually:

• Vertical ascent to Earth escape
• A direct ascent launch profile
• The direct ascent lunar mission architecture

The first, which is what the OP describes, has never been planned by any space agency for any mission. It's not feasible at all. You never simply burn straight up to escape velocity; you always perform a gravity turn. It's certainly possible to do, given enough stages, but it would take so much more fuel that it's absolutely never considered.

The second, the direct ascent launch profile, simply means that your upper stage burns straight through LEO insertion into the BLEO transfer burn without SECO and without any coast period. This was more often done a long time ago, when we didn't have restartable upper stages. It can technically be slightly more efficient due to the Oberth effect but it requires perfect timing and often is not possible depending on your launch site latitude and the inclination and phasing requirements of the mission. And there's still a brief period when you're technically in LEO, even though you never stop burning.

The third describes a mission architecture for going to the moon and coming back. When we were doing our planning for crewed moon missions, there were three options. In a Direct Ascent mission, a single ginormous rocket would take off, go into a LEO parking orbit, burn to the moon, enter lunar orbit, descend to the surface, return to lunar orbit, and then return back to Earth. This never required any orbital rendezvous or docking, but it meant the largest possible launch vehicle. In an Earth Orbit Rendezvous mission, many small launches to LEO would assemble a moon vehicle which, once completed, would burn to the moon, enter lunar orbit, descend to the surface, return to lunar orbit, and then return back to Earth. This could be completed without building any ginormous rocket. In a Lunar Orbit Rendezvous mission, a single ginormous rocket would take off, go into a LEO parking orbit, burn to the moon, and enter lunar orbit, THEN send a small lander down to the surface while the main vehicle remained in lunar orbit, after which the lander would return to the main vehicle and the main vehicle would return home. This meant the original launch vehicle, while still very large, wouldn't have to be nearly as large as in a Direct Ascent approach, and this is what became the Apollo architecture.

All three things can be called "direct ascent" but they are all very different.

Spoiler

It should be noted that there are actually more than just those three options. Here's a (nearly) exhaustive list:

• Monolithic Direct Ascent. As above, you have one vehicle that leaves Earth, lands on the moon, and returns, shedding stages all the way there and back. This was the original plan for the Apollo missions, although it would have required the C-8 Nova rocket with nearly twice the takeoff thrust of the Saturn V. This is also the successful mission architecture for the (robotic) Luna 16 lunar sample return.
• Monolithic Lunar Orbit Rendezvous. You have one vehicle which leaves Earth and enters lunar orbit, but then sends a separate lander down for the landing mission. This offers a huge advantage over Direct Ascent because you leave the re-entry capsule and all of the return-to-earth propellant parked up in lunar orbit instead of dragging it down to the lunar surface and back up, although it does require two separate life support systems. This was the architecture used by the successful Apollo program, and it was the architecture planned for the N-1 lunar landings.
• Earth Orbit Rendezvous. You use a series of smaller launch vehicles to build a larger mission vehicle in low earth orbit, and then that vehicle performs a Direct Ascent mission. This was the architecture favored by Von Braun and (if memory serves me correctly) Buzz Aldrin. It would have taken ten Saturn 1 launches: one for the capsule, one for the ascent-and-return stage, two for the descent stage, one for the lunar orbital insertion stage, and a whopping five to fully refuel the upper stage of the last Saturn 1 to perform the TLI burn. This required an extremely high launch cadence and complex orbital rendezvous and assembly with cryogenic propellant transfer, all of which is possible today but would have been highly aspirational in the 1960s. This is technically also the mission plan for SpaceX Starship Mars missions, where Starships will be retanked in LEO, then go to Mars, land, and then return directly in one piece (albeit with ISRU to retank on Mars).
• Dual Orbit Rendezvous. Same as Lunar Orbit Rendezvous, but you launch your lander vehicle, your return vehicle, and (potentially) your TLI stage separately and assemble them in LEO. This allows you to send much more payload to the moon than a monolithic LOR, with much fewer launches and less complexity than a direct ascent EOR. This was the planned architecture for Constellation, which would have had crew and capsule launching on Ares I while the lander launched on Ares V, using the Ares V upper stage as the transfer stage.
• Distributed Lunar Orbit Rendezvous. Just like Dual Orbit Rendezvous, except that the crew capsule and the lander are each launched directly to cislunar space without ever meeting up in LEO. This requires (at least) two separate transfer burns to TLI and two separate lunar orbit insertion burns, but it allows you to take nearly as much payload as Dual Orbit Rendezvous with smaller launch vehicles. This is the plan for the Artemis missions, except that the SLS-launched crew capsule will rendezvous with the commercially-launched lander in NRHO rather than LLO. With Lunar Starship acting as the commercial lander, this adds on retanking missions in LEO which technically makes it both Dual Orbit Rendezvous and Distributed Lunar Orbit Rendezvous; if NASA had gone with National Team then it would have been a straightforward Distributed Lunar Orbit Rendezvous.
• Sustained Integrated Architecture. This describes any long-term sustainable architecture intended to foster multiple repeat missions to and from a destination using the same system of vehicles. For example, the Space Transportation System (which shrunk and shrunk to eventually become the Space Shuttle and nothing more) originally envisioned a reusable winged chemical shuttle providing regular service between Earth's surface and an LEO space station, a reusable space-only nuclear shuttle launched on Saturn V providing regular service between an LEO space station and a lunar orbit space station, and chemical tugs (brought to the LEO space station by the winged chemical shuttle and then sometimes transferred to the lunar orbit space station by the nuclear shuttle) to launch deep space probes or act as lunar lander stages. With this approach you have dedicated vehicles performing each leg of the journey, all being regularly resupplied through the overall integrated system. That's the future we all want but are unlikely to get.

 

7 hours ago, RCgothic said:

A rocket that could accelerate at 1.5G straight up could accelerate at ~2.3G horizontally whilst maintaining the vertical thrust necessary to counter gravity.

Indeed. It's counterintuitive and it honestly feels a little magical, but Trigonometry Works In Mysterious Ways.

[sevenperforce]

1 hour ago, sevenperforce said:

It should be noted that there are actually more than just those three options. Here's a (nearly) exhaustive list:

• Monolithic Direct Ascent. As above, you have one vehicle that leaves Earth, lands on the moon, and returns, shedding stages all the way there and back. This was the original plan for the Apollo missions, although it would have required the C-8 Nova rocket with nearly twice the takeoff thrust of the Saturn V. This is also the successful mission architecture for the (robotic) Luna 16 lunar sample return.
• Monolithic Lunar Orbit Rendezvous. You have one vehicle which leaves Earth and enters lunar orbit, but then sends a separate lander down for the landing mission. This offers a huge advantage over Direct Ascent because you leave the re-entry capsule and all of the return-to-earth propellant parked up in lunar orbit instead of dragging it down to the lunar surface and back up, although it does require two separate life support systems. This was the architecture used by the successful Apollo program, and it was the architecture planned for the N-1 lunar landings.
• Earth Orbit Rendezvous. You use a series of smaller launch vehicles to build a larger mission vehicle in low earth orbit, and then that vehicle performs a Direct Ascent mission. This was the architecture favored by Von Braun and (if memory serves me correctly) Buzz Aldrin. It would have taken ten Saturn 1 launches: one for the capsule, one for the ascent-and-return stage, two for the descent stage, one for the lunar orbital insertion stage, and a whopping five to fully refuel the upper stage of the last Saturn 1 to perform the TLI burn. This required an extremely high launch cadence and complex orbital rendezvous and assembly with cryogenic propellant transfer, all of which is possible today but would have been highly aspirational in the 1960s. This is technically also the mission plan for SpaceX Starship Mars missions, where Starships will be retanked in LEO, then go to Mars, land, and then return directly in one piece (albeit with ISRU to retank on Mars).
• Dual Orbit Rendezvous. Same as Lunar Orbit Rendezvous, but you launch your lander vehicle, your return vehicle, and (potentially) your TLI stage separately and assemble them in LEO. This allows you to send much more payload to the moon than a monolithic LOR, with much fewer launches and less complexity than a direct ascent EOR. This was the planned architecture for Constellation, which would have had crew and capsule launching on Ares I while the lander launched on Ares V, using the Ares V upper stage as the transfer stage.
• Distributed Lunar Orbit Rendezvous. Just like Dual Orbit Rendezvous, except that the crew capsule and the lander are each launched directly to cislunar space without ever meeting up in LEO. This requires (at least) two separate transfer burns to TLI and two separate lunar orbit insertion burns, but it allows you to take nearly as much payload as Dual Orbit Rendezvous with smaller launch vehicles. This is the plan for the Artemis missions, except that the SLS-launched crew capsule will rendezvous with the commercially-launched lander in NRHO rather than LLO. With Lunar Starship acting as the commercial lander, this adds on retanking missions in LEO which technically makes it both Dual Orbit Rendezvous and Distributed Lunar Orbit Rendezvous; if NASA had gone with National Team then it would have been a straightforward Distributed Lunar Orbit Rendezvous.
• Sustained Integrated Architecture. This describes any long-term sustainable architecture intended to foster multiple repeat missions to and from a destination using the same system of vehicles. For example, the Space Transportation System (which shrunk and shrunk to eventually become the Space Shuttle and nothing more) originally envisioned a reusable winged chemical shuttle providing regular service between Earth's surface and an LEO space station, a reusable space-only nuclear shuttle launched on Saturn V providing regular service between an LEO space station and a lunar orbit space station, and chemical tugs (brought to the LEO space station by the winged chemical shuttle and then sometimes transferred to the lunar orbit space station by the nuclear shuttle) to launch deep space probes or act as lunar lander stages. With this approach you have dedicated vehicles performing each leg of the journey, all being regularly resupplied through the overall integrated system. That's the future we all want but are unlikely to get.

Because I apparently have too much time on my hands, here are some handy little diagrams showing each of the different modalities. Obviously nothing is to scale.

Spoiler

 

 

 

 

[wumpus]

19 hours ago, Superfluous J said:

This is not strictly true.

You can make a rocket for which it's true, but for each of those you can make a cheaper one that goes to orbit first.

Depends on what you call "orbit".  In KSP you aren't in "orbit" unless both your Ap and Pe are above the atmosphere.  An optimal launch would follow a trajectory similar to an orbital trajectory but never stop burning until it got to the final escape trajectory.  It could conceivably do the burn entirely "not in space", but it would take remarkably high thrust to do this under the Kármán line (maybe some of those multistage solid rocket orbital spaceships could pull it off with a small cargo).  Even harder to do so under Virgin Galactic's "definition".  Apollo did a few orbits in a very low parking orbit, but I'm sure that was significantly higher than the Kármán line.

[sevenperforce]

3 hours ago, wumpus said:

Depends on what you call "orbit".  In KSP you aren't in "orbit" unless both your Ap and Pe are above the atmosphere.  An optimal launch would follow a trajectory similar to an orbital trajectory but never stop burning until it got to the final escape trajectory.  It could conceivably do the burn entirely "not in space", but it would take remarkably high thrust to do this under the Kármán line (maybe some of those multistage solid rocket orbital spaceships could pull it off with a small cargo).  Even harder to do so under Virgin Galactic's "definition".  Apollo did a few orbits in a very low parking orbit, but I'm sure that was significantly higher than the Kármán line.

I believe Apollo parking orbits were around 100 nautical miles or around 186 km, which is definitely well above the Karman line.

Technically, you are always in an "orbit" of some kind unless you are physically touching the ground. It's just that some portion of your orbit (and, usually, most of your orbit) has a nasty tendency to intersect the planet, which can prove extremely exciting, and not in a good way. A perfectly-optimal launch would reach an apogee at some point high in the upper atmosphere and burn through that apogee to convert it to a hyperbolic perigee with the desired escape trajectory.

[magnemoe]

You save dV doing an direct burn over getting into low earth orbit like Apollo. And as I read most missions do this an Apollo was an exception. 
Or if you want an inclination who don't fit with your launch window? 

But you always want an gravity turn. you only burn upward to not crash and to get our of atmosphere , its wasted, that count is horizontal velocity. 

[farmerben]

It's certainly not wasted velocity if you don't come back down.  Everybody should try this in KSP you might be surprised how well it works.   This is anecdotal, I find that a rocket powerful enough to orbit on the first stage can easily reach apogee millions of km above Kerbin.  The second payload stage from there can easily go on it's interplanetary mission, without wasting much second stage fuel.   But I can stage early enough to still get good Oberth effect with the second stage.  Launching from dawn I usually vector a little bit west after I get out of the atmosphere to get the best escape trajectory vector, which I know is inefficient, but not too much

The trig losses are not large once you get far out from the surface of the planet.  The higher thrust the better (this may make it unrealistic for human flight).  The first stage boosters can go as high as you like and come straight back down, offset by the planets rotation.  Which is good for first stage powered landing.  Especially when we in real life there is no first stage with thermal shielding enough to survive reentry from orbit.  Powered vertical descent with several burns could handle it.  

[darthgently]

21 minutes ago, farmerben said:

It's certainly not wasted velocity if you don't come back down.  Everybody should try this in KSP you might be surprised how well it works.   This is anecdotal, I find that a rocket powerful enough to orbit on the first stage can easily reach apogee millions of km above Kerbin.  The second payload stage from there can easily go on it's interplanetary mission, without wasting much second stage fuel.   But I can stage early enough to still get good Oberth effect with the second stage.  Launching from dawn I usually vector a little bit west after I get out of the atmosphere to get the best escape trajectory vector, which I know is inefficient, but not too much

The trig losses are not large once you get far out from the surface of the planet.  The higher thrust the better (this may make it unrealistic for human flight).  The first stage boosters can go as high as you like and come straight back down, offset by the planets rotation.  Which is good for first stage powered landing.  Especially when we in real life there is no first stage with thermal shielding enough to survive reentry from orbit.  Powered vertical descent with several burns could handle it.  

I'm fairly certain Oberth would not be a factor in a vertical ascent as your velocity vector is not perpendicular to the body

[magnemoe]

On 10/29/2022 at 4:10 PM, farmerben said:

It's certainly not wasted velocity if you don't come back down.  Everybody should try this in KSP you might be surprised how well it works.   This is anecdotal, I find that a rocket powerful enough to orbit on the first stage can easily reach apogee millions of km above Kerbin.  The second payload stage from there can easily go on it's interplanetary mission, without wasting much second stage fuel.   But I can stage early enough to still get good Oberth effect with the second stage.  Launching from dawn I usually vector a little bit west after I get out of the atmosphere to get the best escape trajectory vector, which I know is inefficient, but not too much

The trig losses are not large once you get far out from the surface of the planet.  The higher thrust the better (this may make it unrealistic for human flight).  The first stage boosters can go as high as you like and come straight back down, offset by the planets rotation.  Which is good for first stage powered landing.  Especially when we in real life there is no first stage with thermal shielding enough to survive reentry from orbit.  Powered vertical descent with several burns could handle it.  

In one way you are correct. One way to take advantage of the centaur upper stage  low dry mass and high ISP but low trust is to launch first stage to an higher Ap than your intended orbit if your target is LEO. 
You then use the time second stage takes getting to Ap and down to circulate the orbit.
This is also why the starliner need an centaur with two engines. If you get an abort the high launch angle and therefore the high decent angle if you get an second stage fail would be bad for the crew. 
Falcon 9 upper stage has much higher TWR so they don't have this issue.  Note that real world rocket don't do circulation burns in LEO like we do in KSP, they just burn second stage until orbit is circular. 
If orbit is high they do but here it might as well be the satellite doing this who is common for GEO. 

[sevenperforce]

On 10/28/2022 at 3:01 PM, magnemoe said:

You save dV doing an direct burn over getting into low earth orbit like Apollo. And as I read most missions do this an Apollo was an exception.

There is a small difference in efficiency between a direct burn and a parking orbit, but it's rarely going to make or break anything. The degree of efficiency improvement is also often a factor of the T/W ratios of the each stage. If the stage which provides the bulk of the Δv for orbit (whether that's a second/sustainer stage which doesn't make orbit or the same stage that performs the BLEO burn) has a low T/W ratio, then it will need to have a lofted trajectory, which results in Oberth losses because you're performing your burn from a higher altitude. This effect is further multiplied because a stage with a higher T/W ratio can perform the BLEO burn more quickly, which means it can take more of an advantage of the Oberth effect.

Other changes in efficiency are associated with any Δv lost to drag in the parking orbit. This is small but not always insignificant; Centaur continually vents hydrogen press gas to make up for drag losses in its parking orbit. If you have an upper stage with a low T/W ratio and so you have an initially lofted trajectory, then there will be a balance to how much of a perigee raise you perform. Too much, and you're losing Oberth efficiency; too little, and you incur higher drag losses during your dip through the very upper reaches of the atmosphere.

On 10/28/2022 at 3:01 PM, magnemoe said:

Or if you want an inclination who don't fit with your launch window? 

Not just inclination; it's also an issue of phasing.

On 10/29/2022 at 10:10 AM, farmerben said:

Everybody should try this in KSP you might be surprised how well it works.   This is anecdotal, I find that a rocket powerful enough to orbit on the first stage can easily reach apogee millions of km above Kerbin.  The second payload stage from there can easily go on it's interplanetary mission, without wasting much second stage fuel.

Sure, it works. Anyone who plays around with KSP will probably have experienced this. But just because you CAN do something doesn't mean it's the most efficient way of doing it. For any rocket design, you will find that you can deliver greater payload to an interplanetary orbit by entering a parking orbit first, rather than launching straight up, if you perform a reasonably efficient gravity turn.

On 10/30/2022 at 12:25 PM, magnemoe said:

Falcon 9 upper stage has much higher TWR so they don't have this issue.  Note that real world rocket don't do circulation burns in LEO like we do in KSP, they just burn second stage until orbit is circular. 

Because Earth is physically much larger than Kerbin, orbital velocity is much higher, and upper stage T/W ratios tend to be lower, your orbital insertion burn is MUCH longer than it would be on Kerbin. Accordingly (at least generally speaking), you're much less likely to end up with an ascent profile where your time to apogee is very long at all.

If you've practiced launching from the Mun, you really get good at aiming for an optimal ascent where your time-to-apogee is less than 10-20 seconds all the way until you've completed circularization.

The Space Shuttle and SLS are good examples of a common exception. Because they are sustainer architectures, the T/W ratio is quite high at burnout and so the time to apogee increases rapidly at the end of the burn, leading to the need for suborbital tank staging and a long circularization burn.