Guide for optimizing staging

[Tyko]

Hey all, I'm trying to get better at optimizing stages during a launch. Are there some guidelines, or (yikes) math,  that others use to decide how much DV each stage should contribute based on the engines and mass?

[Pecan]

1 hour ago, Tyko said:

Hey all, I'm trying to get better at optimizing stages during a launch. Are there some guidelines, or (yikes) math,  that others use to decide how much DV each stage should contribute based on the engines and mass?

A lot depends on the mass the engines can lift, since you have to keep the appropriate TWR in each stage but as a first approximation roughly equal dV.  The next questions are ... what does 'appropriate' TWR mean and is it vacuum or sea-level ISP that counts when calculating dV?  Anything above about 3km you can use vacuum ISP.  Launch needs at least 1.3 TWR (I find) and each stage doesn't really benefit from more than 2.  Higher stages can get away with lower TWRs since they're mostly building speed but not altitude.

So: Pick an upper-stage engine for its vacuum ISP.  Load it with (the payload, obviously) and as much fuel as it can lift.  Pick the next stage engine for its sea-level thrust and see if it can lift the upper-stage to at least 3km and contribute the dV to orbit that the upper stage lacks.  (If the upper stage doesn't lack then you've got a SSTO!).  If that doesn't work change the engine for vacuum ISP and make it a mid-stage, then repeat for the initial launch stage or strap a couple of SRBs on it.  If you need more than 3 stages to launch your rocket is too big :-)

[A35K]

Pretty much what has been said, if you want some specific Dv numbers I can tell you this: In general I tend to have about 1.2-1.5 km/s of Dv on the 1st stage (including any side boosters), which is a powerful atmospheric optimized engine and this should be enough (with appropriate TWR as explained by @Pecan), to get you to about 40-45 km apoapsis with a speed of 600-800m/s, then you fire the second stage, a vaccum optimized one with about another 1500m/s dv and that should get you safely in LKO. 

Edit: Obviously don't start turning just when you fire the second stage, you should start way before that. That speed figure I gave for the first stage should be a significant horizontal component, not purely vertical

Edited January 4, 2019 by A35K

[Snark]

Launchpad TWR for the first stage should generally be in the range of 1.3 to 2.0.  (The very broad range is due to differing philosophies of rocket design.  You can build a perfectly good launcher that lifts at 1.3, or one that lifts at 2.0-- but they'd look pretty different and have quite different ascent profiles.)

I don't look at dV per stage-- I look at mass ratio.  Specifically, what's the ratio of fuel mass to engine mass in a stage?  I aim to keep that somewhere in the 5-10 range (higher numbers for upper stages that need lower TWR).

 

[Tyko]

here's an example of what I'm trying to understand...I often have to throttle a stage's engines during a launch so I'm not raising my AP too fast during a gravity turn. If I'm running a bunch of engines throttled at 30% for a long time i'm clearly losing efficiency because I'm lifting a bunch of dead weight. On the other hand, I'm also lifting dead weight if I'm staging too often, so sometimes it makes sense to run in a throttled down state for a bit if it saves me a stage - the Saturn V famously turned off the middle F-1 engine mid-launch to lower thrust but continued to carry the heavy engine. So there are times when it makes sense to run with throttled / shut down engines rather than staging. I'm just trying to figure out how long is too long for running in that throttled mode.

 

[bewing]

Throttling is almost never useful -- it almost always wastes much more dV than it saves. Just so long as your gravity turn is reasonably aggressive, you are almost always much better off burning at full throttle until your Ap is where you want, then coasting to Ap, then raising your Pe above the atmosphere. Circularizing is usually a mistake too.

If, by the time your ship reaches 40km, your Ap is not high enough -- then crank your nose all the way over to the horizon and keep burning at full throttle.

Edited January 4, 2019 by bewing

[Snark]

23 minutes ago, Tyko said:

I often have to throttle a stage's engines during a launch so I'm not raising my AP too fast during a gravity turn.

If your AP is doing that, it means your gravity turn needs to be way more aggressive.  Unless you have a very unusually draggy ship design, you should be running at 100% throttle continuously from liftoff until your Ap gets to where you want it.

What TWR do you usually launch with, and how many degrees east do you usually tip (i.e. immediately on liftoff) to start your gravity turn?  What angle is your ship above horizontal when you reach 300 m/s?

Also, I'm confused, what do you mean "raising AP too fast"?  You mean you're climbing too steeply?

[Tyko]

10 minutes ago, Snark said:

If your AP is doing that, it means your gravity turn needs to be way more aggressive.  Unless you have a very unusually draggy ship design, you should be running at 100% throttle continuously from liftoff until your Ap gets to where you want it.

What TWR do you usually launch with, and how many degrees east do you usually tip (i.e. immediately on liftoff) to start your gravity turn?  What angle is your ship above horizontal when you reach 300 m/s?

Also, I'm confused, what do you mean "raising AP too fast"?  You mean you're climbing too steeply?

my normal flight profile is to aim for 45 degrees at 10km. 

I'll launch upwards to somewhere between 50 and 100m/s, tip 10 degrees and then lock SAS to prograde. Once I hit a time to AP of 50 seconds I'll throttle the engines to hold me at 50 seconds. Then I'll just ride prograde at 50 seconds to AP all the way to orbit, staging as necessary.

This works pretty well and gets me consistent launches, the question is just always when the 1st stage should be dropped. One thought is that I should be staging right when my time to AP hits 50 seconds and the next stage should be TWR balanced to hold me at 50 seconds as the gravity turn continues. This means I'm never throttling those big booster engines and not carrying their dead weight, but there's also an argument for keeping them around a bit longer so the 2nd stage doesn't need as much power. This is the conundrum.

[Snark]

1 hour ago, Tyko said:

my normal flight profile is to aim for 45 degrees at 10km.

Oh.  Yeah, that's an incredibly inefficient ascent profile.

"Go straight up 10 km and then crank it 45 degrees" used to be the standard advice back in the old "souposphere" days before KSP 1.0, and it was good advice... in 2015.

Since then, however, it's really outdated.  KSP 1.0 completely changed things around.  You should start your gravity turn immediately off the pad.  Immediately upon liftoff crank it N degrees eastwards and set SAS to hold , then just burn until your Ap gets up to where you want it.

"Okay, so how many degrees is N?" I hear you cry.  Well, that depends on your TWR-- if you launch with a low TWR like 1.3 it'll be barely a degree or two, but if you launch with a higher TWR like 2.0 (as I do) it'll be close to 10 degrees.  With a bit of practice you'll get fairly good at estimating it.  The sanity check is to see how fast you're going when you get 45 degrees off the vertical.  It ought to be around 350-400 m/s.  For a high TWR like 2.0, that means you'll hit 45 degrees and 400ish m/s when you're still only six or eight km altitude.  If you're lower TWR, then that point will be a bit higher up... but if you reach 12 km altitude and you still haven't tipped over at least 45 degrees from the vertical, you're not turning hard enough.

Try that on for size-- it will save you quite a lot of dV getting to orbit.  (As in, "hundreds of m/s", I'm guessing.)

[EDIT] Hmmm, upon reading what you wrote, I realize that maybe what you meant was "I start my gravity turn immediately, and I aim to be traveling 45 degrees off vertical by 10 km altitude", in which case your ascent profile is already pretty much what I was describing.  In which case I have no idea why you're throttling down, or what you mean when you think your AP is "rising too fast".  Why wouldn't you burn at 100% until Ap is where you want it, in order to minimize gravity losses?

[Tyko]

12 minutes ago, Snark said:

Oh.  Yeah, that's an incredibly inefficient ascent profile.

"Go straight up 10 km and then crank it 45 degrees" used to be the standard advice back in the old "souposphere" days before KSP 1.0, and it was good advice... in 2015.

Since then, however, it's really outdated.  KSP 1.0 completely changed things around.  You should start your gravity turn immediately off the pad.  Immediately upon liftoff crank it N degrees eastwards and set SAS to hold , then just burn until your Ap gets up to where you want it.

"Okay, so how many degrees is N?" I hear you cry.  Well, that depends on your TWR-- if you launch with a low TWR like 1.3 it'll be barely a degree or two, but if you launch with a higher TWR like 2.0 (as I do) it'll be close to 10 degrees.  With a bit of practice you'll get fairly good at estimating it.  The sanity check is to see how fast you're going when you get 45 degrees off the vertical.  It ought to be around 350-400 m/s.  For a high TWR like 2.0, that means you'll hit 45 degrees and 400ish m/s when you're still only six or eight km altitude.  If you're lower TWR, then that point will be a bit higher up... but if you reach 12 km altitude and you still haven't tipped over at least 45 degrees from the vertical, you're not turning hard enough.

Try that on for size-- it will save you quite a lot of dV getting to orbit.  (As in, "hundreds of m/s", I'm guessing.)

[EDIT] Hmmm, upon reading what you wrote, I realize that maybe what you meant was "I start my gravity turn immediately, and I aim to be traveling 45 degrees off vertical by 10 km altitude", in which case your ascent profile is already pretty much what I was describing.  In which case I have no idea why you're throttling down, or what you mean when you think your AP is "rising too fast".  Why wouldn't you burn at 100% until Ap is where you want it, in order to minimize gravity losses?

Snark...I didnt say I went straight up...never said that...I appreciate you're trying to help, but please read...

here's the second line again "I'll launch upwards to somewhere between 50 and 100m/s, tip 10 degrees and then lock SAS to prograde. Once I hit a time to AP of 50 seconds I'll throttle the engines to hold me at 50 seconds. Then I'll just ride prograde at 50 seconds to AP all the way to orbit, staging as necessary."

Again, I appreciate you're trying to help, but don't read my first sentence which clearly states I AIM to be at 45 degrees and take that to mean I TURN to 45 degrees

EDIT: yes, you didn't read it LOL...the reason I'm throttling down is because I'm trying to keep my time to AP at 50 seconds so i'm continually dropping into a more circular orbit. If I keep the thrust at 100% then my time to AP keep climbing and the ascent gets steeper not more flat. This means that I spend more DV doing my circularization at AP. By keeping time to AP constant at 50 seconds my typical circularization burn is about 25m/s. if I go a full on burn to AP my circularization cost is has been 80 to 100 DV, that's a big difference

 

Edited January 5, 2019 by Tyko

[bewing]

1 minute ago, Snark said:

"Go straight up 10 km and then crank it 45 degrees" used to be the standard advice back in the old "souposphere" days before KSP 1.0, and it was good advice... in 2015.

Ah, snark -- you jumped to a conclusion on that one.  That's not what he said. He said he goes up straight until 50 to 100 m/s, then cranks it over 10 degrees and then verifies that it's at 45 degrees at 10km.

I find myself that if prograde reaches 45 degrees at 8km, that's more like it. So 10 may be a little high.

 

[Snark]

3 minutes ago, Tyko said:

Snark...I didnt say I went straight up...never said that...I appreciate you're trying to help, but please read...

Yup, thus my comment, which you quoted,

10 minutes ago, Snark said:

[EDIT] Hmmm, upon reading what you wrote, I realize that maybe what you meant was "I start my gravity turn immediately, and I aim to be traveling 45 degrees off vertical by 10 km altitude", in which case your ascent profile is already pretty much what I was describing.  In which case I have no idea why you're throttling down, or what you mean when you think your AP is "rising too fast".  Why wouldn't you burn at 100% until Ap is where you want it, in order to minimize gravity losses?

^ which is the point where I realized and corrected my goof.

And also asked the question "In which case I have no idea why you're throttling down, or what you mean when you think your AP is "rising too fast".  Why wouldn't you burn at 100% until Ap is where you want it, in order to minimize gravity losses?" which has still not been answered.  So apparently I'm not the only person who's not reading everything... 

Seriously though, can you explain the reasoning, i.e. why throttling down?

4 minutes ago, bewing said:

I find myself that if prograde reaches 45 degrees at 8km, that's more like it. So 10 may be a little high.

Depends a lot on what TWR you like to fly with.  For example, I like high TWR launches of about 2.0, so I usually hit 45 even lower than 8 km.  But I could see someone who likes lower TWR getting quite a bit higher before hitting 45.

[Tyko]

13 minutes ago, Snark said:

Yup, thus my comment, which you quoted,

^ which is the point where I realized and corrected my goof.

And also asked the question "In which case I have no idea why you're throttling down, or what you mean when you think your AP is "rising too fast".  Why wouldn't you burn at 100% until Ap is where you want it, in order to minimize gravity losses?" which has still not been answered.  So apparently I'm not the only person who's not reading everything... 

Seriously though, can you explain the reasoning, i.e. why throttling down?

Depends a lot on what TWR you like to fly with.  For example, I like high TWR launches of about 2.0, so I usually hit 45 even lower than 8 km.  But I could see someone who likes lower TWR getting quite a bit higher before hitting 45.

Got it...yep...When I started this thread I thought "OMG...I'm going to have to wade through tons of comments assuming I don't know about the basics of launching." I've been playing for 3 years and am well beyond that.

I really am talking about ways to shave off that last 100 m/s of DV. your method results in a higher circularization cost at AP while mine is burning more on ascent by not using the engines as efficiently, but my circularization is really cheap because I'm circularizing all the way up. If I could build my stages so I'm not having to throttle as much it would be better - hence the reason I started this thread 

 

Edited January 5, 2019 by Tyko

[Snark]

3 minutes ago, Tyko said:

If I could build my stages so I'm not having to throttle as much it would be better

Yeah, throttling down during ascent is pretty much bad by definition-- because it means you're lugging around more engine dead weight than you need to be.  So if there's a stage where  you throttle down, it means you should have less engine on the stage.

What sort of TWR values are typical for your 1st stage, 2nd stage, 3rd stage?  In general you want your 1st stage to have a pretty darn high TWR since it has to do that initial dead lift.  But with a fairly aggressive gravity turn, then you're already tipped fairly far over by the time that first stage burns out, and thus the 2nd stage really doesn't need very much TWR at all-- even under 1 can be okay, since it's traveling mostly sideways at that point, so you can just ride at 100% throttle all the way to the point where your Ap gets where you want it-- and that's at a very shallow trajectory, so the circularization burn is actually quite modest.

Exactly how the balance of dV works out depends on exactly what sort of path and mix of TWR values you're following now, but I'm guessing that you'll probably find this way more efficient in terms of conserving dV.

[bewing]

10 minutes ago, Tyko said:

I really am talking about ways to shave off that last 100 m/s of DV. your method results in a higher circularization cost at AP while mine is burning more on ascent by not using the engines as efficiently, but my circularization is really cheap because I'm circularizing all the way up.

You really should try both methods and compare fuel/dv in orbit. I was doing something similar as an exercise several months ago -- and I found the "circularize at Ap" method to be noticeably more efficient.

[Aegolius13]

I found this thread very informative - in addition to ascent profile stuff, it compares the drag loss of going faster to the gravity loss of taking longer to get to orbital speed.  The conclusion was that, if you're launching anything that's halfway aerodynamic, you want to go as fast as you can (without blowing up), and in as flat of a trajectory as possible.  So that would suggest it's not advisable to throttle back or turn off engines.  

 

As you say, there are advantages to staging off weight that's no longer needed.  But throwing away a whole set of engines in favor of a different set (like Saturn V) is not terribly efficient either, since it means the second-stage engines are dead weight at liftoff.   Asparagus staging offers a bit of a cheat here versus real life, offering some of the benefits of both staging and not staging. 

Very generally, I tend to use two staging approaches when getting to orbit.  I have no math to support these, just heuristic experience suggesting they work well. They also tend to be decent from a debris-management perspective.

1.  "Space Shuttle" - Start with a core engine and a couple radial boosters (either SRBs or asparagus-staged liquid boosters).  Plan to get to orbit with some fuel left in the core stage, or (at worst) ditch it around the time of making orbit, as with the STS.  The boosters should obviously be optimized for sea level, but there is some flexibility with the core.  Sustainer-type engines like the Swivel, Skiff and Skipper work, or even something like the Rhino, that's not mediocre (but not useless) at sea leave. [EDIT - "Space Shuttle" does not imply carrying the payload on the side of the core.  I just meant a similar parallel-stage-to-orbit engine configuration.]

2. "Falcon 9" -  Launch a single stack with a decent sea-level engine (Skipper-ish or better), and stage that off in favor of a vacuum engine partway through ascent.  Plan to have quite a bit of delta-v left in the second stage upon reaching orbit.  

  

 

Edited January 5, 2019 by Aegolius13

[FleshJeb]

@Tyko Scroll to the Optimization section:

http://www.projectrho.com/public_html/rocket/multistage.php

I think this is outdated for some engines, but you can check your work against it:

https://garycourt.github.io/korc/

(I really need to start playing 2.5x or 3.2x, because I have no compelling reason to optimize launches at 1x scale.)

[Tyko]

@Snark & @bewing - thanks to both of you for the advice. I'm still figuring out optimal staging, but I did a ton of testing last night and have a few results you might be interested in.

1. tip-over earlier is definitely better - The earliest you can tip for your TWR and still keep your time to AP going up is definitely more efficient. In my test rocket with a middlin' TWR of 1.35 the difference between tip-over at 65 m/s versus 100 m/s was over 100 DV
2. Tipping a small amount at 10-20m/s didn't really improve over doing a fast 10 degree tip-over a bit higher up. I settled on doing a 10 degree tip as a standard and then just adjusting the speed at which I tipped. This was a little less fiddly than trying to do a tiny tip-over right at launch and didn't noticebly affect my DV results
3. This is most interesting - there wasn't much of a DV difference between throttling to a steady 50 m/s time to AP and going full-bore all the way up. There was a fairly linear trade-off between DV spent on ascent vs DV spent circularizing at AP. Going full-bore made for a faster ascent which, as a player, is kinda nice. Throttling took longer to get to altitude, but resulted in less atmospheric heating - several times with full-bore ascents the front of my craft overheated.

As far as staging it seemed like the best option was to run the first stage for as long as possible - either full-bore or throttling. Even if I was throttling, any inefficiency of carrying the heavy, throttled engine when past 45 degrees was lower than the DV cost of lifting the dead weight of your second stage engine in the first place.

Edited January 10, 2019 by Tyko

[Starhawk]

On 1/4/2019 at 4:46 PM, Tyko said:

Then I'll just ride prograde at 50 seconds to AP all the way to orbit, staging as necessary

This technique has never worked for me.  I always find that I need to steer the craft somewhat during ascent to get the curve right.  The difference between start and end TWR for a given stage and the sudden change in TWR when staging seems to be the biggest reason why I can't just lock prograde.  I never worry about my time to apo as long as it's greater than 45 seconds.

I will second what Snark and bewing said about it never being a good idea to throttle down during ascent.  I noticed that you didn't indicate what TWR you like to use for your lifters.  I find the TWR makes a significant difference to the ascent profile.  As, of course, does heat tolerance.

Anyway, I will always choose to steer the craft rather than throttle down during ascent.

 

Happy landings!

[Tyko]

2 hours ago, Starhawk said:

This technique has never worked for me.  I always find that I need to steer the craft somewhat during ascent to get the curve right.  The difference between start and end TWR for a given stage and the sudden change in TWR when staging seems to be the biggest reason why I can't just lock prograde.  I never worry about my time to apo as long as it's greater than 45 seconds.

I will second what Snark and bewing said about it never being a good idea to throttle down during ascent.  I noticed that you didn't indicate what TWR you like to use for your lifters.  I find the TWR makes a significant difference to the ascent profile.  As, of course, does heat tolerance.

Anyway, I will always choose to steer the craft rather than throttle down during ascent.

 

Happy landings!

Thanks for the advice! I'm not sure if you read the last post I made right above yours - check point #3. I did an evening's worth of testing and talk about what I found.

Throttling during ascent doesn't really affect total orbit DV. Whatever you lose in the slower ascent you make up when you make your final circularization burn. It takes longer to get to orbit, which is a few minutes of wasted time, but you can leave SAS locked on Prograde and just control the throttle all the way up, so you're not fighting as much with the controls. it's also pretty forgiving if you miss your initial tip-over point a bit - I got tired of restarting takeoffs over and over trying to the tip-over point just right.

Edited January 18, 2019 by Tyko

[Starhawk]

Well, I did read that point and wasn't quite sure why that would be the case, so I set it aside in my mind before writing my reply.

Upon further reflection I realize that my blanket statement about it never being a good idea to throttle down is wildly inaccurate.
I should have said that it is never a good idea to throttle down when you are ascending directly upward.  And it's less of a good idea the closer you are pointed to upward.
OTOH, the more horizontal your heading the less it matters if you throttle down (other than spending longer in atmo).

So, if you're throttling back later in the ascent I believe it doesn't make any significant difference in gravity losses.

I definitely understand the convenience factor of locking SAS and just worrying about the throttle.

Also, I see you mentioned a TWR of 1.35 (which I originally missed).  To me this is a very low TWR for launch.

Anyway, your ascent method is much more of a true gravity turn, whereas I just force my craft into what seems like an efficient trajectory.

Happy landings!

[OhioBob]

I use this...

 

[Superfluous J]

Basically, launch your ship and, the moment you want to throttle down, check how much fuel you have left in that stage. Then, go back to the VAB and use that knowledge to figure out how much dV your first stage used, up to the point where you wanted to throttle down.

That's the dV that your first stage should have. Your second stage should ideally have 3300 minus that amount, and a TWR equal to what you wanted to throttle down to.

Note this could mean dropping the entire first stage, or dropping side boosters leaving only the center engine to have that low TWR.

[Tyko]

@Starhawk & @OhioBob

Thanks for the ideas and for the link. I'm out of "likes" for the day   

1 hour ago, Starhawk said:

Also, I see you mentioned a TWR of 1.35 (which I originally missed).  To me this is a very low TWR for launch.

I'm playing with 30% science returns in a 2.5x (or 3.2x) scaled system. When you have limited engine tech and 5000-6000 DV to orbit it's tough to pull off higher TWR. Also 1.3-1.6 is "realistic" for 20th century rockets.

• American rocket stacks had pretty low TWR
  • Saturn V: 1.18
  • Gemini/Titan II: 1.24
  • Mercury/Atlas: 1.26
• Russian craft, with the 4 huge side boosters, had higher TWRs
  • Vostok: 1.65
  • Soyuz:1.6

Edited January 18, 2019 by Tyko

[Tyko]

41 minutes ago, 5thHorseman said:

Basically, launch your ship and, the moment you want to throttle down, check how much fuel you have left in that stage. Then, go back to the VAB and use that knowledge to figure out how much dV your first stage used, up to the point where you wanted to throttle down.

That's the dV that your first stage should have. Your second stage should ideally have 3300 minus that amount, and a TWR equal to what you wanted to throttle down to.

Note this could mean dropping the entire first stage, or dropping side boosters leaving only the center engine to have that low TWR.

Thanks! I tried that and it's not always worth it. Here's what I found in testing:

If you have side boosters that you're discarding - like STS, Soyuz, Falcon Heavy, Delta IV Heavy - your approach works great because you're shedding tank and engine mass at just the moment you don't need them anymore.

If you're using traditional stack it's a different story though. When you're on initial ascent, when weight matters the most, you're fighting gravity with the dead weight of a second stage engine, which really saps efficiency. On the other hand, if you keep your first stage longer into the flight and throttle it you're also carrying some dead mass because you're not using the full potential of the engine, but you're doing so when you're already into your gravity turn, most of your thrust is horizontal and weight isn't as much of a concern.

I subscribed to your approach originally and it wasn't until I did some actual side by side testing that I discovered that, in a stack, keeping the first stage longer is actually a good idea. I'm still trying to figure out when it makes sense to stage. Using as few stages as possible seems like the best plan. If you can SSTO there's no reason to add a second stage. There's a point of no return in which the first stage can't lift enough fuel to SSTO and this is when additional stage(s) needs to be added.

 

 

 

Edited January 18, 2019 by Tyko