considerations for coast length when going for maximum altitude?

[Drew Kerman]

Let's keep the problem simple - you have a 3-stage rocket (two boosters + payload) shooting straight up that you want to get as high as possible. Your first thought would be to just fire both boosters in sequence, but then you're traveling too fast and working in a thicker atmosphere. Your second thought would be to fire the first booster, separate and wait until the second booster reaches its peak altitude and fire it off, but then you're working with less speed even tho the ISP is better as the engine is weaker.

So where is the sweet spot between the two? I know it depends greatly on the rocket specs and I didn't give those, but surely there is some general theory to put to work here?

[Aegolius13]

Generally you want to burn as early as possible, to minimize gravity losses and maximize Oberth.  At least until the point your rocket is burning up or going out of control.

This thread has some good stuff:

 

Edited November 18, 2016 by Aegolius13

[Streetwind]

When you are launching straight up, as your simplified example goes, you want zero coast time as long as you still have fuel and engines remaining. As a rule, aerodynamic losses are vastly overestimated by almost every player, perhaps partly in thanks to the extremely visible mach and shock heating effects KSP features. If my ship is on fire going up, they reason, the game must be telling me that I'm doing it wrong. But that is not so. On a normal gravity turn launch, about 70% of your dV goes towards orbital velocity, about 26% goes to gravity, about 3% goes towards aerodynamic losses, and the last missing percent goes towards too much manual steering. It's really, really difficult to intentionally reach a situation where aerodynamic losses ever actually matter.

I've personally tested this "launch straight up" scenario you speak of a few times, out of pure curiosity. Every single time, the highest TWR vehicle had the highest achieved altitude. That included vehicles exceeding TWR 5 on the pad and 15 near the end of their burn - that still wasn't enough to make aerodynamic losses matter more than gravity. I have never succeeded in building a test vehicle for this experiment that had too much thrust. Not even when the burn was really short and focused on the lower atmosphere.

Of course, your test vehicle must stay stable after burnout. If your rocket flips at the end because after losing all fuel the engine is now the heaviest part, then you're going to get massive induced drag. Aerodynamic losses are so low because rockets are pointy. If they go engine-first, or even sideways, that is no longer given. In such a case, you'll instantly slow down a great deal.

Edited November 18, 2016 by Streetwind

[Padishar]

8 hours ago, Drew Kerman said:

So where is the sweet spot between the two? I know it depends greatly on the rocket specs and I didn't give those, but surely there is some general theory to put to work here?

This is a classic rocket science problem (google for Goddard problem).  There was a fun challenge thread about this back in 1.0.4

Basically, you want to always keep accelerating upwards but it can improve your altitude to throttle back through the transonic region.  It only takes a very small difference in speed when your engines burn out to make a considerable difference to your final altitude...

[Palaceviking]

I've just really started using pre-made lifters(subassemblies) and have found that staying as close to prograde as possible, continuous burning and my gravity turn ending exactly at desired apoapsis allows me to lift some really heavy stuff with a disgustingly low twr.

1.anytime spent coasting is incurring gravity losses.

2.the continuous burn may sometimes put my apoapsis too high due to either poor piloting or desperation to get out of the atmosphere but any 'fall' after this just adds to my circularisation energy and reduces the burn.

3. not 100% sure but i think in reality continuous burns are standard.

4.I find a starting twr of 1.15(for gods sake don't turn until 100m/s)and a second stage twr of between 0.3 and.06 seem to work best for this method.

5.   3% to aerodynamic losses is at best still 99m/s. That's handy once your orbital. 

Edited November 18, 2016 by Palaceviking

[Snark]

The general rule is "burn as hard as you can, as early as you can."  Unless you have an insanely high TWR or a very awkward, draggy rocket design, then you'll always want to floor it as hard as you can and keep it that way.

Yes, there's a potential concern about aerodynamic drag losses, but:

7 hours ago, Streetwind said:

As a rule, aerodynamic losses are vastly overestimated by almost every player, perhaps partly in thanks to the extremely visible mach and shock heating effects KSP features.

^ This.  Yes, it's technically possible to build a rocket that's so overpowered (i.e. has a sky-high TWR) that it does, in fact, need to throttle back a bit to avoid losing too much to aero.  However, in practice this very rarely comes up.

Here's the deal:

• You want to minimize gravity losses, which means "go fast soon".
• You want to minimize aero losses, which means "don't go fast soon."

Since these two goals are mutually contradictory, it's a matter of "which one is more important".  The faster you go, the less the gravity loss and the greater the aero loss.  So, where's the "sweet spot" that provides optimum efficiency during climbing?

Well, it turns out that the answer is, the most efficient vertical ascent is when you're climbing at terminal velocity.  In other words, you want your rocket to be climbing at the same speed that it would fall if it were just plummeting straight down with no engine power.

There may be some grizzled KSP veterans out there who remember the pre-1.0 "souposphere" days of the old aero, who may be thinking "now hold on there a minute, sonny, the 'climb at terminal velocity' thing was an OLD-aero thing, and stopped being true when we went to the new aero."

Actually, that's not the case.  "Climb at terminal velocity" was true before 1.0, and is still true post-1.0.  The change to aerodynamic drag calculation didn't affect that fundamental principle.  What it changed was, just what IS the terminal velocity.  Used to be that it was a simple number based on altitude, regardless of ship design-- and it was really low, so you needed to care about it a lot.  (Terminal velocity at Kerbin sea level was only around 100 m/s.)  Now it depends on the size and shape of your rocket, and in general tends to be a lot higher than it was-- especially if you have a rocket that's reasonably streamlined.

And the terminal velocity gets higher in a hurry as the atmosphere gets thinner.

What this means, in practical terms, is that unless you have an insanely high TWR or a very ungainly/draggy rocket design, your rocket never "catches up" to its terminal velocity-- it simply doesn't have the oomph.  There's a good chance that a well-designed rocket already has a terminal velocity of 200 m/s or more, just sitting on the launch pad; and by the time it has accelerated to 200 m/s, it's at an altitude where the terminal velocity is already considerably higher than 200 m/s, and so forth.

In other words:  your rocket is always moving slower than terminal velocity, which means it's always better to be going faster than you are, which means burn full-throttle right off the pad and keep it that way.

[Foxster]

As above ^^ 

Going as quick as you can without heatploding will get the job done these days. 

[Snark]

One last comment-- the "floor it all the way" advice above is generally true for launching from Kerbin.  If you're doing an Eve ascent, then depending on ship design, exceeding terminal velocity may be more of a practical concern.  Terminal velocity is considerably lower than on Kerbin (yes, the gravity's 70% more, but the atmosphere is five times thicker), and you have to climb vertically for a really long way before you can start the gravity turn.

So, for Eve ascent, not only is it easier to accidentally exceed your terminal velocity, but also there's a bigger penalty for ascending at significantly above or below that speed (since the ascent lasts so long).

Knowing what your terminal velocity is, is really hard, since that varies both with altitude and with the shape of your ship.  A slightly easier (though still nontrivial) thing you can do is, try to keep your local TWR right at 2.0, at least during the first 15 km or so of your ascent.  If your ship's local TWR is 2 during a vertical ascent, then it will stick pretty close to terminal velocity.  Of course, your TWR changes as you burn fuel and as you tweak the throttle, so that's not exactly a walk in the park, either-- but at least you can see what your ship mass and your engine thrust are, whereas knowing your actual terminal velocity is tricky, short of some specialized mod, or running an actual experiment.

...All of which just boils down to "ascent from Eve is a bear," but then, you already knew that. 

[foamyesque]

Oh, hey, I remember an old-school Goddard problem thread from waaaay back in the day. Posted by Doc Evo, as I recall. That was fun... though I always preferred the drag races (i.e, minimum time to a specified altitude). Somewhat counterintuitively, with those, you did want long coast times simply because it let you get a higher TWR earlier -- and an extra m/s a second after launch is worth a lot more than an extra m/s a second before you cross the altitude marker.

 

Might be interesting to resurrect that, adjusted for KSP's new atmosphere.

[Padishar]

22 minutes ago, foamyesque said:

Might be interesting to resurrect that, adjusted for KSP's new atmosphere.

Back in the day, you got a higher altitude from a pod and hammer if you turned the thrust limiter down.  This is no longer the case, it really is quite difficult now to go fast enough for the drag to make a difference.  Aerodynamic stability is the main problem with going fast now, e.g. the vessel in the Goddard thread I linked above will flip once the engine burns out if you launch at full throttle and the extra drag will kill the altitude, but throttling down before mach 1 and keeping the throttle low until you get above 10k before flooring it again will get you out of the atmosphere now where it was very hard to get above 62k back in KSP 1.0.4.

Adding a bit more payload mass, e.g. a few extra stack batteries, should reduce the altitude it gets to and may make it less flip happy.  But this may also result in full throttle getting a higher altitude so it may need more drag to keep it interesting...

[foamyesque]

Just now, Padishar said:

Back in the day, you got a higher altitude from a pod and hammer if you turned the thrust limiter down.

You and I have different ideas of "back in the day", I think. :v