A question about upper stage design to LEO or GTO

[Gaarst]

(This is about RSS but the only thing that changes are dV requirements, not gameplay)

So I design all my launchers for LEO (9.3km/s of dV) which means that most of them have rather large and powerful upper stages (compared to real rockets I mean, TWR is still usually below 1) since they are meant to finish the orbital burn (give most of the horizontal velocity to the rocket).

For the sake of curiosity (and because I like building stuff more than launching it), I recently started to work on upper stages separately from the rest of the lifter. While so far i just took a small upper stage and stick it on top of the LEO lifter, I have started to integrate the dV needed for a GTO burn inside the launcher itself, to have rockets specifically designed for GTO. Said rockets usually have 3 smaller stages instead of the usual 2 large I use.

But for some reason I realised that putting a lighter payload on a big lifter to achieve the 11.something km/s dV for GTO was more efficient than building the whole rocket for GTO. Similarly, I designed a rocket for GTO but it is actually worse at putting things in GTO than other launchers I have that were designed for LEO, and better at putting things in LEO. So clearly I'm doing something wrong.

Is the best way to achieve high GTO efficiency to just put a small upper stage on a lifter (like Proton-M/Briz-M), put a lighter payload on a large lifter and launch it directly for GTO (Falcon 9) or to design the whole thing for GTO (Ariane 5) ? And is a large upper stage that finishes the orbital burn + the GTO insertion burn better than a smaller stage that is only lighted once in space?

Edited November 24, 2016 by Gaarst

[Streetwind]

It may look like a complex problem at first glance, but ultimately it can be reduced to something very simple:

- Any rocket going to GEO, goes to LEO first.
- Therefore, any payload going to GEO does so by coming from LEO.
- Therefore, the mass a lifter can send to GTO or GEO is closely related to just two things: the mass it can send to LEO, and the rocket equation.

Your goal, therefore, is to place into LEO the most efficient transfer stage you can build that has enough dV to insert your payload into GTO or GEO (depending on whether the payload can finish the insertion itself or not). "Efficient" here means the specific impulse of the engine and the dry mass of tankage plus engine. The transfer stage can be the same upper stage that finished LEO insertion, or it could be a dedicated transfer stage. The downside of the former is that you need to lug the whole second stage to GTO, and the downside of the latter is that you need to lift the additional hardware for the extra stage to LEO.

You can see this in real life: a Falcon 9 running a heavy kerosene upper stage can put 22,800 kg into LEO, and 36.4% of that (8,300 kg) into GTO. An Ariane 5 ECA running a small hydrogen upper stage can put 21,500 kg into LEO, and 48.8% of that (10,500 kg) into GTO. Less weight to LEO than the Falcon 9, but more to GTO! That 33% difference is solely located in the efficiency of the upper stage, which is poor on the F9. It uses a less efficent fuel, and it needs to be built extra big and powerful because it has to allow the first stage to cut out much earlier than other rockets to facilitate recoverabiliy (even though the payload numbers given are for expendable mode). The Ariane 5, on the other hand, uses very large boosters in a "stage and a half" concept to let the first stage burn extra long, requiring only a tiny, lightweight upper stage for finishing LEO insertion and going to GTO.

Edited November 25, 2016 by Streetwind

[Gaarst]

2 hours ago, Streetwind said:

It may look like a complex problem at first glance, but ultimately it can be reduced to something very simple:

- Any rocket going to GEO, goes to LEO first.
- Therefore, any payload going to GEO does so by coming from LEO.
- Therefore, the mass a lifter can send to GTO or GEO is closely related to just two things: the mass it can send to LEO, and the rocket equation.

Your goal, therefore, is to place into LEO the most efficient transfer stage you can build that has enough dV to insert your payload into GTO or GEO (depending on whether the payload can finish the insertion itself or not). "Efficient" here means the specific impulse of the engine and the dry mass of tankage plus engine. The transfer stage can be the same upper stage that finished LEO insertion, or it could be a dedicated transfer stage. The downside of the former is that you need to lug the whole second stage to GTO, and the downside of the latter is that you need to lift the additional hardware for the extra stage to LEO.

You can see this in real life: a Falcon 9 running a heavy kerosene upper stage can put 22,800 kg into LEO, and 36.4% of that (8,300 kg) into GTO. An Ariane 5 ECA running a small hydrogen upper stage can put 21,500 kg into LEO, and 48.8% of that (10,500 kg) into GTO. Less weight to LEO than the Falcon 9, but more to GTO! That 33% difference is solely located in the efficiency of the upper stage, which is poor on the F9. It uses a less efficent fuel, and it needs to be built extra big and powerful because it has to allow the first stage to cut out much earlier than other rockets to facilitate recoverabiliy (even though the payload numbers given are for expendable mode). The Ariane 5, on the other hand, uses very large boosters in a "stage and a half" concept to let the first stage burn extra long, requiring only a tiny, lightweight upper stage for finishing LEO insertion and going to GTO.

Thanks for your answer! I'll play around with hydrolox, and see if I can get better efficiency.

[foamyesque]

It is also not necessarily true that anything going to GTO or GEO goes through LEO first; you can launch straight into a suborbital transfer to GTO. That has implications (which aren't a problem in stock KSP, but could be important with realism overhauls) concerning your engine starts and stuff like that.

[Gaarst]

21 minutes ago, foamyesque said:

It is also not necessarily true that anything going to GTO or GEO goes through LEO first; you can launch straight into a suborbital transfer to GTO. That has implications (which aren't a problem in stock KSP, but could be important with realism overhauls) concerning your engine starts and stuff like that.

I mean more in terms of dV than trajectory. I don't literally stop at LEO and restart my engines there, I take a steeper trajectory and burn until I get an apo at 36000km. I just build my rockets with 9.3km/s for LEO and put whatever I need on top for GTO.

[MaxL_1023]

I usually build my rockets from the top down. Start with the payload you want to put into GTO, including whatever engines it has. In this case, I assume that the GTO-GEO would be done by the satellite itself unless I have an efficient, restartable engine available (think RL-10 or Aestus II). 

I then pick an engine about 4-5 times more powerful than whatever I would need to have a TWR of about 0.8 if it was just the payload. I then stick fuel tanks until I get a TWR of 0.8 with the new engine. I repeat this (4-5 stage to stage thrust ratio) with a TWR of 0.8 for anything less than about 2000 m/s short of orbit, 1.0 for a stage igniting in atmosphere, and 1.25 on the pad. The results usually work fine for me. 

I normally make a custom launcher for every payload, since the "no restarts" for many otherwise excellent engines means a lighter payload would just waste the extra fuel you can carry. Later in the tech tree you can get away with this kind of generic launching.

[Streetwind]

11 hours ago, foamyesque said:

It is also not necessarily true that anything going to GTO or GEO goes through LEO first; you can launch straight into a suborbital transfer to GTO. That has implications (which aren't a problem in stock KSP, but could be important with realism overhauls) concerning your engine starts and stuff like that.

True! But the issue with that is that it's inefficient. You'll get higher gravity losses, and a huge orbital insertion burn that requires an oversized engine and benefits less from the Oberth effect.

Since @Gaarst is asking for a way to improve his efficiency of launching GEO payloads, that kind of trajectory is not really applicable as a solution to the problem.

[foamyesque]

15 minutes ago, Streetwind said:

True! But the issue with that is that it's inefficient. You'll get higher gravity losses, and a huge orbital insertion burn that requires an oversized engine and benefits less from the Oberth effect.

Since @Gaarst is asking for a way to improve his efficiency of launching GEO payloads, that kind of trajectory is not really applicable as a solution to the problem.

 

Nah, it's actually more efficient and benefits more from the Oberth effect. You might be confusing a vertical launch with a direct insertion; the two are not the same. A direct insertion executes a gravity turn and so on, just as an insertion into a low orbit does, but instead of stopping with an apoapsis of 100km (or whatever), you leave the engines running and run your apsis up to geostationary altitude in one shot. You get an Oberth advantage because you're burning your engines lower down and with a higher velocity than you would be going from a parking orbit, and you save on deltaV since you cut out the unnecessary circularization burn in low orbit.

 

EDIT:
 

You're correct that you need a decent TWR to capture the full benefits, though, since as the burn gets longer you start getting more of a vertical component and/or start having issues with precision, whereas an orbit-to-orbit transition can make use of apsis kicking. Most chemical engine'd transfer stages should be entirely capable of it, but if you're using an ion engine or some other high efficiency, low TWR, unlimited start engine the parking approach could be more efficient.

Edited November 26, 2016 by foamyesque

[Streetwind]

5 minutes ago, foamyesque said:

You might be confusing a vertical launch with a direct insertion; the two are not the same.

In that case, though, we can return to the simplification I made in my first post, which already included this sort fo trajectory. A direct insertion into GTO ends up with an orbit that has a periapsis in LEO, or a mere handful of m/s away from it. It flies the same trajectory as a rocket that goes to LEO, except that it doesn't stop its engines and continues to push the apoapsis further.

Therefore: "Any payload that goes to GEO, goes to LEO first."

Edited November 26, 2016 by Streetwind

[foamyesque]

It's been quite awhile since I played with RSS -- the amount of mods needed to reinstall after game updates has turned me off -- but in stock KSP it's entirely possible to hit a Kerbin-Mun trajectory with a negative periapsis, and to complete your burn to do so somewhere around 40km.

[GoSlash27]

Gaarst,

 

On 11/24/2016 at 2:11 PM, Gaarst said:

(This is about RSS but the only thing that changes are dV requirements, not gameplay)

Actually, a lot more than just the DV requirement changes. Having no experience with RSS, I have no idea where staging becomes more mass efficient wrt DV budget. In the stock game that's generally around 2,300 m/sec. In RSS... I have no idea. You have different Isps, engine masses, tank mass ratios, etc.

 I'd guess (emphasis on "guess") that squeezing another 2km/sec out of a (roughly) 5km/sec transstage would be lighter overall than tacking one on in RSS. This would not be the case in stock.

Best,
-Slashy

 

[GoSlash27]

As for the whole "you have to go through LEO" argument, that's pedantry. You do not have to establish LEO before moving on, but you *do* have to expend the DV it would've required along the way. The difference in total DV is negligible and not worth arguing about IMO. For planning purposes, it's fine to assume circ in LEO before pressing on, even if you're not actually going to do that.

Edited November 26, 2016 by GoSlash27

[foamyesque]

22 minutes ago, GoSlash27 said:

As for the whole "you have to go through LEO" argument, that's pedantry. You do not have to establish LEO before moving on, but you *do* have to expend the DV it would've required along the way. The difference in total DV is negligible and not worth arguing about IMO. For planning purposes, it's fine to assume circ in LEO before pressing on, even if you're not actually going to do that.

 

That's sort of true -- the dV difference isn't extreme -- but one of the things the RSS/RO stuff does is impose limits on engine starts, which impacts design on things like stretching your upper stage vs. adding a third stage; either can do a direct insertion, but if you can't re-light, only a third transfer stage can do the GTO injection.

[GoSlash27]

6 minutes ago, foamyesque said:

 

That's sort of true -- the dV difference isn't extreme -- but one of the things the RSS/RO stuff does is impose limits on engine starts, which impacts design on things like stretching your upper stage vs. adding a third stage; either can do a direct insertion, but if you can't re-light, only a third transfer stage can do the GTO injection.

foamyesque,

  All Gaarst wants to know is which approach is more mass efficient. He has already demonstrated that he is capable of doing it either way, so he's clearly aware of these technical points.

Best,
-Slashy

[Gaarst]

1 hour ago, GoSlash27 said:

Actually, a lot more than just the DV requirement changes. Having no experience with RSS, I have no idea where staging becomes more mass efficient wrt DV budget. In the stock game that's generally around 2,300 m/sec. In RSS... I have no idea. You have different Isps, engine masses, tank mass ratios, etc.

 I'd guess (emphasis on "guess") that squeezing another 2km/sec out of a (roughly) 5km/sec transstage would be lighter overall than tacking one on in RSS. This would not be the case in stock.

You're right, but TBH, I was mostly writing this to avoid it being moved to Add-on Discussions where it would have been forgotten.

As for mass efficiency of staging it's mostly a matter of fuel, you're going to have a hard time pulling more than 5km/s out of an hypergolic stage while a hydrolox would reach that with one tank.

For the squeezing part, it's kind of a non-issue. Having more than 5km/s of dV on your upper stage would mean that it needs to give most of the initial horizontal velocity (the hardest part of a launch in RSS since if you don't do it right, you'll end up reentring 3km/s slower than orbital velocity) so you'd either need a powerful engine to get that velocity fast enough, adding mass and decreasing efficiency; or you'd have terrible TWR at your upper stage ignition which means you'd have to take a steeper, less efficient trajectory. EIther case, your stage will be huge.
For an ascent stage, you definitely want to add a stage after 5km/s of dV, it might be heavier but it will end up being more efficient. For an orbital stage adding tanks is better.
You don't end up with more than 5km/s ascent stages (neither do real rockets which typically have small upper stages) anyway for LEO. As for GTO, that's the reason I've made this thread.

34 minutes ago, foamyesque said:

That's sort of true -- the dV difference isn't extreme -- but one of the things the RSS/RO stuff does is impose limits on engine starts, which impacts design on things like stretching your upper stage vs. adding a third stage; either can do a direct insertion, but if you can't re-light, only a third transfer stage can do the GTO injection.

I do all launches, LEO or GTO, with a single burn from the lifter. The handful m/s dV required for circularisation are done by the orbital stage or payload which anyway are designed to be ignited several times. Adding a third stage or not is not a matter of being able to restart or not, in my case.

Edited November 26, 2016 by Gaarst

[GoSlash27]

3 hours ago, Gaarst said:

For an ascent stage, you definitely want to add a stage after 5km/s of dV, it might be heavier but it will end up being more efficient. For an orbital stage adding tanks is better.

Gaarst,

 If this is the case, then it appears you'd be better-off running 3 stages instead of 2. I mean... it takes your upper stage almost 5km/sec just to get to orbit, right?

If you're not realizing the full potential, I'd guess that it's because you're trying to design from the bottom up rather than the top down.

Good luck!
-Slashy

Edited November 26, 2016 by GoSlash27

[septemberWaves]

I don't know how much help this might be since I play with the stock scale planets, but generally if I'm going to place something in stationary orbit above Kerbin, I'll use a light vacuum stage beneath the payload that has enough delta-v to move from my usual parking orbit into stationary orbit, and then place both the payload and that vacuum stage on top of a lifter that can efficiently transport their combined mass to the 80km parking orbit. I imagine some rockets use a similar method in reality (though I am not certain about this), and it's probably a fairly useful technique for RO as well since it doesn't require designing a new launch vehicle.

[Gaarst]

1 hour ago, GoSlash27 said:

 If this is the case, then it appears you'd be better-off running 3 stages instead of 2. I mean... it takes your upper stage almost 5km/sec just to get to orbit, right?

If you're not realizing the full potential, I'd guess that it's because you're trying to design from the bottom up rather than the top down.

It's not that simple because on a typical 2 stage design, the first stage is big and the upper stage very small. I find that splitting dV evenly is not very efficient, so I typically use large first stages, with boosters to give it a kick if needed, with a small upper stage giving at most 3-4km/s dV depending on what's below, even less for 3 stages designs. In this case extending the upper stage isn't much of a problem anyway (and very easy with hydrolox).

1 hour ago, eloquentJane said:

I don't know how much help this might be since I play with the stock scale planets, but generally if I'm going to place something in stationary orbit above Kerbin, I'll use a light vacuum stage beneath the payload that has enough delta-v to move from my usual parking orbit into stationary orbit, and then place both the payload and that vacuum stage on top of a lifter that can efficiently transport their combined mass to the 80km parking orbit. I imagine some rockets use a similar method in reality (though I am not certain about this), and it's probably a fairly useful technique for RO as well since it doesn't require designing a new launch vehicle.

But I love designing rockets (more than I like flying them TBH). Sticking an upper stage on a LEO lifter is what I used to do, but I've started designing launchers specifically for GTO and I'm seeking the most efficient method.

Edited November 26, 2016 by Gaarst

[GoSlash27]

53 minutes ago, Gaarst said:

It's not that simple because on a typical 2 stage design, the first stage is big and the upper stage very small. I find that splitting dV evenly is not very efficient, so I typically use large first stages, with boosters to give it a kick if needed, with a small upper stage giving at most 3-4km/s dV depending on what's below, even less for 3 stages designs. In this case extending the upper stage isn't much of a problem anyway (and very easy with hydrolox).

Gaarst,

 In that case, I don't have any answers for you. The best recommendation I can give you is to analyze the problem using the reverse rocket equation. You will have to populate with parts from RSS instead of stock. Doing that, you can easily see which combination is ideal for your needs. It's a PITA to set up the first time, but it'll make your life a whole lot easier in the long run.

 Good luck,
-Slashy