1.1 and TWR (on the launchpad)

[wumpus]

Back in beta, the souposphere dictated that the most efficient TWR was 2.0 at all times.  This made many people unhappy and was widely regarded as a bad move (on the other hand, it *really* helped that early progression and landings).

With 1.1, the "most efficient" TWR seems to be "how fast can you go without exploding"?  Or at least "how fast can you go while remotely able to make a pitchover" (launching at an angle becomes a real possibility with 1.1.  It was a remarkably effective strategy in 1.0 for smaller rockets such as unmanned satellites, but just might make its own in 1.1).  I've heard that Squad changed the fuel usage wrt to throttle position, but expected that to mean that lower TWRs would be more effective.

With 1.0, players suggested a takeoff TWRs that wildly varied between 1.3-1.8.  I noticed that if you rocket as built in the VAR had a TWR>2.0, that limiting the TWR to 2.0 tended to balance aero and gravity losses best.  On the other hand, a "proper gravity turn" was essentially impossible.  You had to do everything you could just to try to control the rocket before hitting trans- and hyper-sonic travel, and after that just let it ride out of the atmosphere (which wouldn't take long at those speeds).  IMPORTANT: just because reducing a rocket with a TWR>2.0 to 1.6 (or something) makes the craft less efficient (assuming you can wrangle it into orbit without huge cosine losses) doesn't mean that you can't make a rocket with TWR of 1.6 more efficient than the 2.0 TWR rocket.  More thrust requires more engines, and the more powerful engines almost always weigh more than the lighter ones, giving you more dry weight.  High TWR rockets are typically that way due to using SRBs for the first stage.  If your are limited by weight on the launchpad, they are unlikely an option.

Much of my 1.0 testing started with "sounding rockets" (fire a rocket straight up and find out what apoapsis the rocket reaches).  That data was then used for a more useful "launch into orbit" (when it became obvious that control was an issue and that mechjeb (of the time, haven't tried since) was hopeless at launching high TWR rockets).  With 1.1 I don't yet have Kerbal Engineer (is it ready?  I had a notice for KAS but not KE) so I really didn't bother with exact TWR and simply fired my sounding rockets.  However, my known "higher than 2.0" rocket used for simple launches into orbit worked best with full thrust (~2.1) and decreased after that (and had a ap of ~70km).  I then built a presumably higher TWR with a kickback and had the same results.  Trying with liquid rockets (and replacing the capsule with a probe body) showed that TWR should be increased to just below the point of explosion (and to also expect that fairings are more important than in 1.0, but possibly only if you take advantage of use of extra thrust.  This test rocket simply hoped that a nosecone would sufficiently occlude the rest of the rocket (with a gap between the octoprobe and the next fuel tank), but that seemed to get hotter than the nosecone).

To put it bluntly, I was unable to find any TWR that could be reduced to achieve a higher ap.  Presumably the "ideal" TWR (when going straight up) is right below that which causing a heating explosion.  Any launch sideways (to orbit) can be expected to have even greater heating, so I would expect the point of lowering thrust for aero losses would remain much higher than lowering thrust to avoid explosions.  Control issues are expected to get even worse than 1.0 (especially if using TWR>2.0) and I expect to run a bunch of tests launching kickback-based satellites (at least once I have KE and can easily determine TWR).

[blowfish]

This thread is a little old, but should still be relevant.

It's important to note that what you mean by "most efficient" depends on the answer.  If you mean least delta-v, then yes, a higher TWR will be best up until drag losses start becoming significant.  If it's least cost, then the calculation is different and you want a lower TWR.

[wumpus]

Note that it wasn't a matter of optimizing around "getting to orbit with the least delta-v", but more a matter of "how to get the most of an existing rocket" (comparing TWR between different rockets to determine efficiency is foolish.  It is hardly something you design around, with the possible exception of landers).  I've noticed more than a few times where players would "dial back" rockets beyond what was reasonable for KSP>1.0, without checking to see if the rocket could be launched with a higher TWR.  It seems that for KSP=1.1.x, there is even less reason to "dial back" any rocket that doesn't explode on the way up.

If you like to use SRBs on your first stage, it hasn't been uncommon to expect to reduce the thrust of some of those SRBs  (obviously, if you have a liquid engine boosted by SRBs, you dial that down first.  My testing makes even that suspect).  The current edition of KSP appears to have a high cost in efficiency for dialing back those rockets.

[OhioBob]

I haven't tested every scenario but, for the most part, I agree that the only reasons to ever "dial back" a rocket are to (1) maintain control, and (2) prevent overheating.  Other than that, if you've got it, use it.

That being said, I subscribe to the school of though that small engines and low TWR are better than large engines and high TWR.  However, when I'm forced to select an engine that delivers a higher TWR than I would normally choose, I run it at as high a throttle setting as I can.  For instance, I prefer a liftoff TWR in the 1.3-1.5 range.  I obtain this TWR by packing on as much fuel as the engine can handle.  However, if I have a rocket that will deliver a liftoff TWR of, say, 1.8, I would never throttle it back to 1.5 just to get it inside my desired range.  I would only throttle back if I'm flipping out of control, can't make my gravity turn, or overheating.  Lowering TWR by using a smaller and cheaper engine is a good thing, lowering TWR by arbitrarily throttling back is a bad thing because it needlessly increases the gravity losses.

 

 

[bewing]

wumpus: my testing agrees with everything you said. In recent versions, thrust-limiting your SRBs has not been useful, unless you wanted to lower your speed and altitude for some reason (filling a contract, maybe).

The only additional caveat being that the faster a rocket goes in the atmosphere, the harder it is to steer (if you need to steer it).

From a physics perspective, the current system makes perfect sense to me. At higher velocities, the engines are able to do more work (<- using the technical definition of "work").

[Plusck]

Hmmm. I'm a bit torn on the issue.

Often, adding a fairing would not be feasible because it would add a lot of weight and require a redesign, potentially even increasing the drag if the payload is an odd shape.

Going diagonally through the lower atmosphere is also not an option because of overheating and massive drag.

Therefore you have to "dial back" to gain altitude. You can't go off prograde because that would cause severe instability, therefore you are obliged to head mostly upwards before dialling back to allow the rocket to fall to a decent angle for the main thrust to orbit.

Now obviously, that initial altitude-gaining thrust is mostly wasted. You get 10km up with little horizontal velocity to build upon. This is why it is mostly done with SRBs. However, piling on additional fuel also seems to be pointless - you will merely be burning that fuel to support the extra mass of fuel that you never really needed in the first place. You need to burn it because you need to recover a decent TWR for that post-10km phase.

So in effect you end up at 12-15km with exactly the same rocket, at the same velocity and same TWR, as you would have obtained by being heavier from the start and not dialling back the LF engines...

 

Now I'm sure that there is some way of calculating the optimal path between the two options (massively dialling back, or massively piling on fuel), but it's certainly beyond me to try.

[OhioBob]

34 minutes ago, Plusck said:

However, piling on additional fuel also seems to be pointless - you will merely be burning that fuel to support the extra mass of fuel that you never really needed in the first place.

Adding propellant increases the mass ratio, which increases the rocket's Δv.  By lowering the TWR we will increase gravity losses, but overall it is a net gain.  That gain can then be used to carry a bigger payload.  And since the added propellant is cheap, the cost per ton of payload goes down.

For example, let's say we have a rocket that can both deliver 3400 m/s and can reach orbit using 3400 m/s.  We now add on propellant to the point that the rocket can deliver a Δv of 3600 m/s, but in doing so we increase the losses to the point that it now takes 3500 m/s to reach orbit.  We have a net gain of +100 m/s.  Since we don't need this extra 100 m/s, we can increase the payload mass until we bring the rocket's Δv down to the same 3500 m/s that we need to reach orbit.  Generally speaking, the percent increase in payload size will be greater than the percent increase in launch cost, meaning that the payload unit cost goes down.

 

[MaxL_1023]

Gravity losses decrease with increasing TWR while air drag losses increase. For every rocket design, there is an optimal launch profile (TWR, stage TWR and gravity turn) which maximizes payload to any particular orbit. In general, taller, heavier rockets can get away with going faster - air drag is not as significant when you have 100 tons behind every square meter of cross section. Overheating is an issue more so with shallower trajectories and lower orbits - your launch TWR won't matter a lot as you will end up burning horizontal at 40km anyways. 

I find a pad TWR of 1.25 works fine for first-stage dominant rockets (3000 m/s from your first stage), with higher TWR values better when you want to use a vacuum engine as soon as possible (think terrier or poodle second stage).

[Speadge]

just check KER for "terminal velocity". keeping this speed on your ascend is nearly the most fuel-efficient way to get out there

[Warzouz]

15 hours ago, blowfish said:

This thread is a little old, but should still be relevant.

It's important to note that what you mean by "most efficient" depends on the answer.  If you mean least delta-v, then yes, a higher TWR will be best up until drag losses start becoming significant.  If it's least cost, then the calculation is different and you want a lower TWR.

That's very true. I've a test rocket that can reach LKO with only 2850m/s (VAC), even without touching any button. It has a TWR of 6 ! (no cheat, no air breather, just regular rocket launched already inclined)

BUT, it's quite heavy, very expensive and has 0 payload. It has no real purpose

Edited May 13, 2016 by Warzouz

[Plusck]

9 hours ago, OhioBob said:

Adding propellant increases the mass ratio, which increases the rocket's Δv.  By lowering the TWR we will increase gravity losses, but overall it is a net gain.  That gain can then be used to carry a bigger payload.  And since the added propellant is cheap, the cost per ton of payload goes down.

For example, let's say we have a rocket that can both deliver 3400 m/s and can reach orbit using 3400 m/s.  We now add on propellant to the point that the rocket can deliver a Δv of 3600 m/s, but in doing so we increase the losses to the point that it now takes 3500 m/s to reach orbit.  We have a net gain of +100 m/s.  Since we don't need this extra 100 m/s, we can increase the payload mass until we bring the rocket's Δv down to the same 3500 m/s that we need to reach orbit.  Generally speaking, the percent increase in payload size will be greater than the percent increase in launch cost, meaning that the payload unit cost goes down.

 

I understand that; my pondering really only relates to that hugely inefficient phase between launch and the time that horizontal acceleration becomes dominant, since 99% of the vertical acceleration during that time is simply lost to gravity and/or to drag.

What triggered these ponderings was a habit I formed, of adding fuel tanks to SRBs to bring my TWR down. I (theoretically correctly) surmised that a high TWR was wasteful and I could increase the dv of my craft by piling on fuel and running on full thrust, rather than throttling back. So when launchpad TWR was too high, I added fuel tanks and lines to the SRBs to give me a higher dv once I reached orbit.

Therefore, between the original craft and the added-tank craft, there is a stage shortly after the SRBs are discarded where the two versions of this craft will be absolutely identical in terms of total mass, fuel load, payload fraction and whatnot.

And what I noticed - and what triggered these musings - was that this "identical craft" moment seemed to occur at more or less exactly the same time for some designs, and gave me almost exactly the same result in terms of remaining dv in orbit - far less than the difference in calculated total dv on the launchpad.

So while I agree that I could have added a touch less fuel and added a touch more payload, and would have ended up with a marginally more "efficient" craft, that margin has often appeared to be vanishingly small, specifically because of that first phase where almost all of the expended energy will end up being lost to gravity.

Which is why I wonder what the maths says about it.

 

edit: in terms of numbers, bearing in mind that we're pulling these out of thin air...

Assuming this hypothetical lifter has a launchpad TWR of 1.8 and a total dv of 3400 m/s. We add fuel to the lifter's first stage (not to the lifter's second stage, we don't want to reduce its TWR at that point) to increase that to 3600 m/s. As you said, with that lower TWR, it takes us 3500 m/s to reach orbit so our net gain is 50%, but since this is the first stage that means an awful lot of fuel. For the same cost we could have bolted on a few Hammers and throttled back even more until they ran out...

Edited May 13, 2016 by Plusck

[Red Iron Crown]

11 hours ago, Speadge said:

just check KER for "terminal velocity". keeping this speed on your ascend is nearly the most fuel-efficient way to get out there

This is outdated advice, hasn't been true since 1.0 changed the aero model. Now it's more efficient to go as fast as you can without burning up. 

[OhioBob]

7 hours ago, Plusck said:

Assuming this hypothetical lifter has a launchpad TWR of 1.8 and a total dv of 3400 m/s. We add fuel to the lifter's first stage (not to the lifter's second stage, we don't want to reduce its TWR at that point) to increase that to 3600 m/s. As you said, with that lower TWR, it takes us 3500 m/s to reach orbit so our net gain is 50%, but since this is the first stage that means an awful lot of fuel. For the same cost we could have bolted on a few Hammers and throttled back even more until they ran out...

My experience is that strap-on solid boosters are not very cost effective.  Not because of the solid motors per se, but because the cost of decouplers and nose cones are disproportionately high in KSP.  A Hammer cost 400 funds, while a TT-38K radial decoupler and aerodynamic nose cone together cost 840 funds.  It seems nuts to me that the accessories cost more than double the motor, but that's the way it is in KSP.  Solid motors are very cost effective if you can use them inline as a first stage, but when strapped on to the side, my experiments have shown me that they actually hurt cost efficiency until we get up to the size of a Kickback.  That being said, I still use them in practice, but generally only when they allow me to get by with a smaller liquid fuelled engine then I otherwise could.

One good way to make a comparison is to compute the total impulse that we get from a pair of Hammers versus adding more propellant.  Each Hammer has 2.81 25 tons of solid propellant.  For the specific impulse, let's use an average of sea level and vacuum values.  Therefore, the total impulse we get from a pair of Hammers is,

2 x 2.8125 x (170+195)/2 x 9.80665 = 10067 kNs

On the other hand, let's say we add a Rockmax X200-8 Fuel Tank, which has 4 tons of liquid propellant.  How much impulse this will deliver depends on the ISP of the engine.  Let's use 295 s, which is an approximate average of the 2.5 m engines most likely be used in this application.  Therefore, the total impulse we get from the extra propellant is,

4 x 295 x 9.80665 = 11572 kNs

We see that we get more impulse from the extra fuel tank, plus it cost only 800 funds versus 2480 funds for the Hammers + accessories.  So as long as the first stage liquid fuelled engine has enough thrust to handle lifting the mass of the extra propellant, we're money ahead by adding more fuel versus adding a pair of small strap-on solids.  We're almost certain to have greater gravity losses by adding the extra fuel tank (due to lower TWR), but we can afford to pay for that.

(edit)

One possible argument in favor of the strap-on solids is that we get to jettison their empty mass earlier in the flight, while we have to carry the inert mass of the fuel tank all the way to the end of first-stage burnout.  This is true, but from the start the empty fuel tank weighs much less than two Hammers.  The dry fuel tank has a mass of 500 kg, while the inert mass of two Hammers + accessories is 1610 kg.  So with the fuel tank we carry less mass for a longer period of time, while with the SRBs we carry more mass for a shorter period of time.  I'm not sure which option works out better.

Edited May 13, 2016 by OhioBob

[G'th]

Are we presuming that we're all operating the same ascent profiles? Are gravity turns even being considered here? 

The thing about going as fast as you can without burning up (Which Im taking here to indicate as exploding, not just generating reentry effects) is that it leaves you with a highly inefficient method of establishing orbit unless your rocket can handle the extreme stresses you'll be putting on it in order to get a proper ascent profile. IE, putting you back in the Kerbal stone ages where you burn to 10k, go straight to 45*, keep burning till you get your desired apo, cut engines and coast until you're ready to circularize. Except now you're just burning until you get your apo up and then you have to either cut your engines or start burning radial in order to keep a fixed apo. When you cut your engines, you've got too much in your early stages and you're losing all the altitude you gained as you coast (its not much, but its still not efficient), and when you start burning radial you lose a lot of efficiency because you're flying upwards on your side, putting a lot of drag on your rocket that you're fighting against.

Maintaining a low TWR through ascent, on the other hand, gives you room to maneuver your rocket properly through the atmosphere, and with a proper turn for your rocket you're effectively minimizing all the losses you'd incur with any other sort of launch method. TL;DR, Duh.

Even on Kerbin, going vertical is cheap, but required. Its the horizontal that costs (but isn't necessarily required on ascent, its only absolutely necessary to establish orbit. After all, the idea of burning straight up to get into space can work, though its horribly inefficient), and going as fast as you can without burning up isn't the answer to efficiently gaining the horizontal velocity you need for orbit. In short, to me it seems as if the fuel costs are being focused on too hard here where its not really necessary.

You can minimize fuel costs very easily if you get the trajectory correct with your combination of engines, as then each stage can be tweaked to be just right for whats needed during that stages portion of the flight. 

My rockets are designed that way, and each stage has a loose min and max TWR that determines when they get staged, and typically you want these to coincide with your dV for each stage. Using a Saturn V style booster (which most of my rockets are based around) as an example, you have a third stage with the payload mounted thats meant for upper atmosphere/vacuum flight, and its TWR will start at ~1.1-1.3. As this stage, max TWR is ignored because in vacuum high TWR is desirable. Engines and fuel amounts are chosen and tweaked to meet the desired starting TWR and the  required dV for the mission.

Second stage, typically for middle atmospheric flight and potentially vacuum flight as well. TWR again starts at ~1.1 and maxes at 1.6-1.7. For cleanup purposes, dV for the stage + that of the first stage is adjusted to be just under the required minimum for LKO, so that the second stage does not become space junk. The starting TWR is fairly strict as this stage typically ignites right around Max Q and what I call the flame barrier (where Kerbin unrealistically starts rendering reentry effects on everything going past Mach 3), and I want the TWR to be low enough so as to not induce too much heat damage to the rocket as I fly into the upper atmosphere as well as to provide better control over the rocket, as at Max Q my rockets typically require a very light touch in order to be flown without breaking up. Too high of a TWR at this stage, and I'll either have stuff exploding (or getting near enough to it), or I'll have any attempt to maneuver the rocket past "ever so slightly" result in a break up. 

First stage, deep atmosphere flight. Essentially to get the rest of the rocket up and moving, and in a position where the following stage can work at its most efficient. Min TWR starts from 1.1-1.4 (depending on the payload. Lighter payloads need lower start TWR, heavier, higher) and caps at ~1.7-1.8. The fuel amounts for 1st and 2nd stage are chosen to split the required dV at a 0.9:1.1 ratio, respectively. All stages are packed with extra fuel not only as ballast to help in achieving the TWR ranges, but also as a contingency in case I mess up the ascent or decide on a higher orbit. (Which I admittedly do quite often)

With the stages designed like this, so long as I fly the rocket properly I'm wasting the relatively minuscule amounts I keep on the rockets extra. If I bothered to crunch the numbers, I could probably get near 100% efficiency as I wouldn't need the ballast fuel.

[OhioBob]

8 minutes ago, G'th said:

Are we presuming that we're all operating the same ascent profiles? Are gravity turns even being considered here? 

I can't speak for others but, in all of my conversation, I'm assuming that we're making an efficient gravity turn.  I find that I usually get my best results when I have a liftoff TWR in the 1.3-1.5 range, and a second stage start TWR in the 1.1-1.3 range.  I'll accept higher or lower TWR if I have too (there are only a finite number of engines to choose from), but I try to keep as close to the desired range as possible.  I don't have a problem with a higher liftoff TWR as long as I can get the rocket to turn and follow a good trajectory.  I don't like highly lofted trajectories that require an early shut down and a large apoapsis burn.  I think the most efficient trajectory is a flat one with a long burn time and a very small apoapsis burn.  The problems I have with high TWR is (1) rocket may resist attempts to turn it, and (2) it's hard to fly a flat trajectory without excessive heating.  This places an upper limit on what I consider to be an acceptable TWR, though I haven't seriously explored what that value is.  As long as I stay close to my 1.3-1.5 target, I know I'm good.  I think the lower limit on TWR is 1.2, below which the gravity losses are just too great.