Don't fully understand TWR

[paul_c]

Recently I've been refining my rocket designs and have been using lower and lower TWR, around 1.2-1.4, with some success - they 'seem' more efficient and still stay flyable, whether this is my getting better at flying though I'm not sure.

You obviously need TWR>1 for a take off (or a lander in vacuum) but the lower the better? Or is say an extreme example of 1.01 too low, because you'll effectively burn a lot of fuel hovering, instead of getting to where you need to go (above 70km for space). If so, what is the 'best' TWR, is there a mathematical solution or is it simply down to look and feel? Does it relate to drag vs desired altitude?

In my mind, it makes sense that you'd not want too much for later on, because you want to put the horizontal speed element nearer to the Ap, which for efficiency is better to be lower (but above 70km to get out the atmosphere), that way you don't have a shutoff then restart - you'd want to continuously burn through the second half of the gravity turn? Otherwise why take a 6 ton 2000kN engine, when you could have taken a 3 ton 1000kN engine up there to be later thrown away?

And for landers....I've aimed at TWR 3 (of the planet/body you're visiting), that way you don't have to do a slow down burn planned quite so far in advance; and have some 'power' to get control/get out of trouble etc, while still having a good control and you also know that 'hover' is around 1/3 throttle.

[Entropian]

You've pretty much got it down.  The "best" TWR would be infinite, as you spend less time burning fuel against gravity, but this is obviously not possible, and with an atmosphere you would burn up pretty fast.  Generally, the higher TWR the better but at some point the engine mass outweighs the benefits, although there is no formula that I know of.  Just play more KSP and you'll get a feel for it pretty fast.  The horizontal component of the orbital insertion is the same way.  Higher TWR?  Start your burn closer to apoapsis.  Lower TWR?  Start your burn farther away.

[Streetwind]

The dV cost to orbit can be broken up into four things: orbital speed, gravity losses, aerodynamic losses, and steering losses. For launching from stock Kerbin with typical TWRs (1.3 to 1.5) and a reasonably pointy rocket, orbital speed is about 65%, gravity losses are 25%-30%, aerodynamic losses are 5%-10%, and steering losses are one or two percent at most.

You cannot reduce orbital speed, but you can reduce the losses. Out of the three, gravity losses are the largest by far.

Gravity losses are constantly incurred whenever a rocket engine is fired not perfectly perpendicular to the gravity well. The more it is firing towards the gravity well, the worse the losses are. Directly at launch, when the vehicle is perfectly vertical, gravity eats up 9.8 m/s worth of dV for every second the engine fires. 9.8 m/s, every single second, just gone. That's why gravity losses are the biggest loss factor.

So how do you reduce it? Well, if you could turn over harder and sooner, you'd be firing your engine further away from the direction of the gravity well sooner, and for longer. Also, if you could just fire your engine for a shorter time, fewer losses would accumulate. And both of these things have one thing in common: they require you to have higher acceleration. In other words: a higher TWR. This is what TWR does for you, beyond having enough of it to lift off; it makes your ascent more efficient by minimizing gravity losses. Which is what @Entropian already correctly pointed out, just with an actual explanation attached

There are two takeaways from this: one, if you lose more dV to additional engine mass than you save in gravity losses with the thrust from those engines, all you did was waste money; and two, TWR largely ceases to matter once you are in orbit.

(I say "largely", because stretching a burn over too large of a section of the circular orbital path starts incurring you significant cosine losses as you thrust off-prograde for minutes on end both ahead of and after passing the maneuver node. And if your thrust gets so low that you have to fly constant-thrust spiral trajectories, like ion engines require you to do IRL, then the cost of transfers can rise drastically - up to 2.4 times that of an ideal Hohmann transfer, given an infinite SoI and infinitessimally low acceleration.)

[AHHans]

First the easy part: if you are already outside the atmosphere - in orbit or on a body without atmosphere - then higher TWR is in theory better. In reality you a) have problems with the precise execution of maneuvers if the TWR is too high and b) you are more interested in dV anyhow, so you usually are quite willing to sacrifice some TWR for more dV. For the low TWR case of orbital maneuvers see @Streetwind's comment about cosine losses during long burns. For landing and take-off in vacuum the TWR can be as high as you want (with the caveats listed above). In the imaginary case of of an infinite TWR (on a perfectly spherical body) you would just burn horizontal directly at take-off, and do a kind of Hohmann transfer from the surface to the orbit.
Which kind of answers your implicit question: no, having an extended coasting phase before circularizing is not necessarily a bad thing.

For the launch from Kerbin (or Eve for that matter) see @Streetwind's answer. Experience has shown that a TWR of 1.3 to 1.7 on the launchpad is usually a good value. A low TWR will lead to more gravity losses, because you need to steer a steeper trajectory to avoid falling back to the ground. If you have a high TWR (especially once you are outside most of the atmosphere) then you'll want a flatter trajectory to reduce the gravity losses in order to reduce gravity losses and to avoid raising your AP more than necessary. That gives rise to the "keep time-to-AP constant" strategy of launching. But you also don't want too high TWR at launch because that will either cause too high aerodynamic losses (if you do a "regular" gravity turn), or too high gravity losses (if you first go more or less straight up to get out of the atmosphere).

I'm sure there is a way to develop a mathematical expression that will give you an optimized TWR for launches from Kerbin, but I don't know if anyone ever bothered to make one. (I know that you would have to pay me real money to do that. )

 

[paul_c]

Many thanks for the replies, I didn't realise gravity losses were quite such a high proportion - I thought a lot more was due to drag (so lower TWR made sense for that). 

[Snark]

6 hours ago, paul_c said:

Recently I've been refining my rocket designs and have been using lower and lower TWR, around 1.2-1.4, with some success - they 'seem' more efficient and still stay flyable, whether this is my getting better at flying though I'm not sure.

You obviously need TWR>1 for a take off (or a lander in vacuum) but the lower the better? Or is say an extreme example of 1.01 too low, because you'll effectively burn a lot of fuel hovering, instead of getting to where you need to go (above 70km for space). If so, what is the 'best' TWR, is there a mathematical solution or is it simply down to look and feel? Does it relate to drag vs desired altitude?

"Reasonable" launchpad TWR is generally in the 1.2 to 2.0 range.  You can fly well, and "efficiently", within that range-- though the design strategy would differ depending on where in that range you fall.

Lower TWR means higher gravity losses, which is less "efficient".  So a higher TWR is "better" in that regard.

But... higher TWR also means bigger engines, which are more expensive than fuel.  A lower TWR rocket can get away with cheaper engines, it's just going to have a bigger launchpad weight and have a lot more fuel to burn.  It's "wasteful" in that sense... but it's not as though KSC is gonna run out of fuel, and fuel is cheaper than engines.

Lower TWR rockets can also get away with being less aerodynamic than high TWR, since they get up to a higher altitude (where the air is thinner) before they get going fast.  So if you have an awkward payload that's difficult to streamline, a low TWR can be useful.

In other words:  both high TWR and low TWR can be a valid design philosophy, you just gotta suit the rocket design to the TWR.

You generally don't want to go any lower than 1.2, because then your gravity losses blow up.  You also probably don't want to go much higher than 2.0, since then aerodynamic losses really go through the roof, and you end up losing more to those than you save in reduced gravity loss.  Also, TWR > 2 generally means you're probably carrying way too much engine, meaning a lot of unnecessary dead weight, which will eat into your dV.

Note that all of the above discussion is for the launchpad.  Upper stages generally have lower TWR, because they're going mostly horizontally and gravity losses aren't much of a thing.

(For my own designs-- which are usually three stages to orbit-- I go for TWR 2.0 for the launchpad, 0.7 to 1.0 for the 2nd stage, and something really low for the 3rd.  Doesn't mean that's some sort of "ideal", it's just how I roll.) 

[paul_c]

My main concern was weight - a high TWR setup needs a big(ger) engine, if I can do it with low TWR then I can often use a lighter (and as a benefit, cheaper too) engine. And the weight saving of course has the multiple benefits of less fuel and less fuel tank dry mass.

My latest rocket had:

TWR: 1.22..................................................................SRBs only.....................599m/s
TWR: 1.48.................................................................liquid 1st stage...........2595m/s
TWR: 1.03 (40km altitude assumed).............liquid 2nd stage...........1834m/s

and that lot got me to Minmus (elliptical orbit) and back.

[vv3k70r]

7 hours ago, paul_c said:

You obviously need TWR>1 for a take off (or a lander in vacuum) but the lower the better?

In amtosphere You have to take some air in front of You for a ride and You pay with Your precious fuel for this.

Steering cost precious fuel so aim when You ar slow and foloow prograde.

8 hours ago, paul_c said:

Or is say an extreme example of 1.01 too low, because you'll effectively burn a lot of fuel hovering

Depend how fast it grows. Fuel You burned dosent add any longer to W.

You would keep TWR on reasonable level until You get to lower density there You do not have to carry lot of air in front of You.

8 hours ago, paul_c said:

Does it relate to drag vs desired altitude?

Yes.

8 hours ago, paul_c said:

that way you don't have a shutoff then restart - you'd want to continuously burn through the second half of the gravity turn?

If You not burning inside atmosphere You wasting fuel You burn previosly.

First stage move You to the altitude and give speed so it could start on 1.3 and burn out around 2 spliting between vertical in horizontal. Second stage could do under 1 because You passed 45deg, You get vertical kick from first and You giving more to horizontal. Third stage is to set You where You would so it should not be too powerfull. Lower and cheaper You would end up (like rescue mission) gentle stagin is needed.

8 hours ago, paul_c said:

Otherwise why take a 6 ton 2000kN engine, when you could have taken a 3 ton 1000kN engine up there to be later thrown away?

When You are in orbit it matters only how efective You are burning fuel, not how fast (in reasonable manner that dosent force You to cirle on spiral). Operating xenon thruster could be boring. There is no reason to take engines for burning in atmosphere when You are leaving it.

8 hours ago, paul_c said:

And for landers....I've aimed at TWR 3 (of the planet/body you're visiting), that way you don't have to do a slow down burn planned quite so far in advance; and have some 'power' to get control/get out of trouble etc

What kind of trouble? When You starting to descend it is done. You fall like a stone on previously aimed position minus drag (in atmosphere overshoot). Set retrogade and reduce Your speed befor smashin' or in case of atmosphere open parachutes when You passing the spot.

8 hours ago, paul_c said:

while still having a good control and you also know that 'hover' is around 1/3 throttle.

You had control before descent, when You started it is done.

36 minutes ago, paul_c said:

And the weight saving of course has the multiple benefits of less fuel and less fuel tank dry mass.

Drop empty tanks?

[paul_c]

Wow there's a lot to work through on the above answer.

Relating to landings, 

1. Unknown terrain, non-precision
2. Known terrain, precision

The first is when I'm visiting somewhere for the first time or exploring a new area of a previously visited body. I'd prefer to land on a flat bit, so some kind of hover/hold off landing/traverse sideways or away from the originally pinpointed spot is needed. Also, its possible to design and build a lander which can cope with (slight) terrain - I did a horizontal lander which happily landed on a 25deg slope.

The second is when I'm bringing in another module for a base station; or doing a rescue mission; or a tourist mission where I've previously landed in that area so I know it. 

More recently I've gotten better at designing and handling landed modules "on the ground" so that they aren't required to land very near an existing base station (and possibly wipe it out).

[paul_c]

As an example in pictures:

15 ton module is 'flown' like a normal vertical rocket to the known flat surface, note it will be deliberately landed a little away from the base, just in case it wipes it out.

It landed and remained stable vertically, but with the planned placement of retractable landing legs, it was able to do a controlled 'fall' to horizontal. Note its about 800m away from the base. Note the edge-placed engines rather than central one, to allow docking port and  expansion. 

The lander/tractor is sent out to collect it. The tractor is about 7 tons, the module is about 15 tons so needs a third pair of wheels (other modules can 'hang off the back' of the tractor and still be dragged around okay).

The lander/tractor has it 'in tow'. Steering is reconfigured (both axles on lander turn same way; module's wheels don't steer).

Nearly there

Some amount of space is needed to swing it around but with the entire ice lake, no real worries

With retractable legs strategically placed on all modules, and angling it from side to side, height above ground can be fine tuned to allow successful docking. This one went on fine. The previous two needed a nudge by a Kerbal on EVA. Note the "2-to-1" adapter on the right, this can be used strategically to align or vary docking height too, if needs be.

 

[Snark]

On 11/26/2020 at 5:20 PM, paul_c said:

My latest rocket had:

TWR: 1.22..................................................................SRBs only.....................599m/s
TWR: 1.48.................................................................liquid 1st stage...........2595m/s
TWR: 1.03 (40km altitude assumed).............liquid 2nd stage...........1834m/s

and that lot got me to Minmus (elliptical orbit) and back.

Well, first and foremost, of course, is that if it works well for you and does what you need it to do, then that's all that really matters. 

That said, I'd suggest that you could do with significantly lower TWR in 3rd stage.  If you've got over 2500 m/s in your 2nd, then that means by the time the 3rd stage kicks in, gravity loss is simply not a thing anymore.  You could have a TWR of 0.5 or below-- even 0.25 would probably be fine-- and save quite a bit on engine weight.

Also:  If your first stage is currently SRBs-only, then depending on how you have them arranged and what engine you have on your 2nd stage, you might be able to squeeze some extra performance out of this, with just a small design tweak, as a sort of "poor man's asparagus."  Details in spoiler, if you're interested.

Spoiler

The technique I describe here will work if your rocket currently meets the following criteria:

• Your liquid-fueled 2nd stage has an engine that's unobstructed by the 1st stage.  For example, if your SRBs are radially mounted on the sides, and you don't have anything attached underneath the 2nd stage LFO engine.
• Your 2nd stage LFO engine has a reasonable atmospheric Isp, i.e. isn't a "vacuum" engine.  (For example, a Reliant or Skipper is okay, a Poodle would not be.)

The idea goes like this:

• Toggle the radial decouplers that attach your SRBs to have crossfeed enabled.  (It's disabled by default.)
• Add a small LFO tank on top of each radial SRB.  Size matters, more on how to gauge this is below.
• Adjust your staging so that you lift off the pad on the SRBs and also the LFO 2nd stage engine.

What happens here is, when your ship lifts off, the LFO engine will be draining fuel out of the lfuel tanks atop the SRBs.  It won't be using its own fuel supply at all.  In this way, it will no longer be dead weight in the 1st stage-- it will be helping right from the get-go.

You choose the size of the little fuel tanks atop the SRBs by doing a little bit of math, so that they run out just before the SRBs burn out.  Look at your LFO engine in the VAB, and the info panel will tell you how many units per second of LFO it burns.  Take that number, multiply by the burn time of your SRBs, and divide by the number of SRBs.  Then pick a tank size that's no bigger than that.

When you're launching, watch those side tanks, and when they drain down to zero, kill the throttle on your LFO engine so that at that point you're just lifting on the SRBs alone (so as not to waste your 2nd-stage fuel yet).

Note that adding your LFO thrust to the SRBs will give you a higher TWR off the pad, which will reduce your gravity losses.    Alternatively, if you still prefer keeping a lower TWR, you could reduce the thrust limiter on the SRBs, which means you could have the same launchpad TWR, but you'd be getting a longer burn time out of the SRBs.  Either way, it's a win.

Basically, a sort of "poor man's asparagus".

 

[vv3k70r]

19 hours ago, paul_c said:

1. Unknown terrain, non-precision

Send a probe. Send a mouse, then a dog, and then our finest. Spot gonna be well known with great precision.

19 hours ago, paul_c said:

I did a horizontal lander which happily landed on a 25deg slope.

In such a case I'm adding some grid structure so after touchdown they can eventually roll. But I didnt get into such an issue and tested if it works after corect landing. Terrains in KSP are very plain even on slopes, it is because of computing power. There is no real rocks, no shifting sands, no "swamps" made of oil or agresive chemistry.

19 hours ago, paul_c said:

More recently I've gotten better at designing and handling landed modules "on the ground" so that they aren't required to land very near an existing base station (and possibly wipe it out).

Navigation in game is extremaly easy so You can land on the spot. I landed manualy inside 100m circle in daylight and around 350 away in dark after 2 landings so I guess it must be much easier for experienced players.

19 hours ago, paul_c said:

As an example in pictures:

Did You try to roll using reaction wheels? They are so powerfull no electric motor from wheels can match it. I suspect negative resistence in wiring^^

19 hours ago, paul_c said:

It landed and remained stable vertically, but with the planned placement of retractable landing legs, it was able to do a controlled 'fall' to horizontal. Note its about 800m away from the base. Note the edge-placed engines rather than central one, to allow docking port and  expansion. 

If You feel a need to keep engines. I rather land a "landing pad" that is docked to base unit that include all electronic (pad have only legs, engines, tank and docks to grab a fuel for another vessel that gonna fly). Simpler version of lunochod solution (without a platform that extend rails for rollout). Similiar to Your design just with this tank and engines left standing on its legs at landing spot.

19 hours ago, paul_c said:

The previous two needed a nudge by a Kerbal on EVA. Note the "2-to-1" adapter on the right, this can be used strategically to align or vary docking height too, if needs be.

Reaction wheels also do this job. But they shouldnt.

You can also level using dampers/springs on wheels.

If we are about TWR it depends mostly on how bad cargo has to travel. Cargo is the problem, not tweaking thrust to propper level. When You left atmosphere TWR refer only to burn time.

 

 

[paul_c]

3 hours ago, Snark said:

Well, first and foremost, of course, is that if it works well for you and does what you need it to do, then that's all that really matters. 

That said, I'd suggest that you could do with significantly lower TWR in 3rd stage.  If you've got over 2500 m/s in your 2nd, then that means by the time the 3rd stage kicks in, gravity loss is simply not a thing anymore.  You could have a TWR of 0.5 or below-- even 0.25 would probably be fine-- and save quite a bit on engine weight.

Also:  If your first stage is currently SRBs-only, then depending on how you have them arranged and what engine you have on your 2nd stage, you might be able to squeeze some extra performance out of this, with just a small design tweak, as a sort of "poor man's asparagus."  Details in spoiler, if you're interested.

  Reveal hidden contents

The technique I describe here will work if your rocket currently meets the following criteria:

• Your liquid-fueled 2nd stage has an engine that's unobstructed by the 1st stage.  For example, if your SRBs are radially mounted on the sides, and you don't have anything attached underneath the 2nd stage LFO engine.
• Your 2nd stage LFO engine has a reasonable atmospheric Isp, i.e. isn't a "vacuum" engine.  (For example, a Reliant or Skipper is okay, a Poodle would not be.)

The idea goes like this:

• Toggle the radial decouplers that attach your SRBs to have crossfeed enabled.  (It's disabled by default.)
• Add a small LFO tank on top of each radial SRB.  Size matters, more on how to gauge this is below.
• Adjust your staging so that you lift off the pad on the SRBs and also the LFO 2nd stage engine.

What happens here is, when your ship lifts off, the LFO engine will be draining fuel out of the lfuel tanks atop the SRBs.  It won't be using its own fuel supply at all.  In this way, it will no longer be dead weight in the 1st stage-- it will be helping right from the get-go.

You choose the size of the little fuel tanks atop the SRBs by doing a little bit of math, so that they run out just before the SRBs burn out.  Look at your LFO engine in the VAB, and the info panel will tell you how many units per second of LFO it burns.  Take that number, multiply by the burn time of your SRBs, and divide by the number of SRBs.  Then pick a tank size that's no bigger than that.

When you're launching, watch those side tanks, and when they drain down to zero, kill the throttle on your LFO engine so that at that point you're just lifting on the SRBs alone (so as not to waste your 2nd-stage fuel yet).

Note that adding your LFO thrust to the SRBs will give you a higher TWR off the pad, which will reduce your gravity losses.    Alternatively, if you still prefer keeping a lower TWR, you could reduce the thrust limiter on the SRBs, which means you could have the same launchpad TWR, but you'd be getting a longer burn time out of the SRBs.  Either way, it's a win.

Basically, a sort of "poor man's asparagus".

 

That's a really clever way to arrange staging - partly because it also makes the reduction in engine throttle, coincide with the increase as each stage's fuel is depleted. I tried it on a heavy(ish) rocket and it worked well.