"Best" TWR Values>

[The Flying Kerbal]

Hey Guys,

So after spending many months just cobbling together rockets on a very much trial and error basis, I've finally started to take building them a bit more seriously.  I've got Kerbal Engineer to help me and have been fiddling about with various Thrust/Weight Ratios to try to make my creations more efficient.

Looking at real world rockets, the Saturn V had an initial TWR of 1.15, whereas the Russians had some of their rockets shooting off at 1.62!  So is there an ideal TWR for a rocket at launch?  I set one up last night which took off with 1.5; it was a multi stage affair lifting a payload into LKO with a lift block, and a stage for a transfer burn to Minmus.  I had all stages set to fire at 1.5, or as close as I could get.  The rocket worked really well, using a Delta V map I was surprised by just how close I came to hitting the Delta V requirements, despite my somewhat iffy piloting ability.

So my questions are, how do you guys figure out what TWR you want your rockets to deliver, and was it a good idea to have each stage set to commence its burn at 1.5?

As a little footnote, the mission went perfectly, right up to when Kerbin's atmosphere was supposed slow the returning vehicle down on its first pass so a stable orbit could be achieved when I raised the Pe back above 70K.  Unfortunately I dropped first the Pe to 35K without thinking, and then the returning vessel (which had no parachutes and a terrier engine) into the ocean at 80m/s!  All the science gathered (and there was a lot) was all lost, but it did make a most impressive splash as it plunged under the waves... 

Edited October 21, 2017 by The Flying Kerbal

[The Aziz]

I typically change the throttle limiter to the setting that give me exactly 1.22 on the first stage. The rest I left untouched. Works quite well with my launch profile (straight up to 100m/s, 23° at 5km, 45° at 10k, 67° at 20km, almost horizontal at 40km)

[MoarBoostersRUS]

I usually try and keep it at about 1.5-1.6 TWR, start heading east at 100m/s and totally horizontal between 40-45km up then ride that to whatever ap im going for. Can get a perfect 150km orbit with 3.3km/s of Dv.

for returns from anything but LKO I have a heat shield especially if there is science onboard drop pe to 35-40km and let it ride and have 2 drogue shutes and a main all radial, with a probe core with reaction torque and a 200ec battery to be safe If it's one of the gray ones, if it's the gold one just put the shutes on top and a shield on the bottom and let it drop in.

Edited October 21, 2017 by MoarBoostersRUS

[Harry Rhodan]

3 hours ago, The Flying Kerbal said:

and was it a good idea to have each stage set to commence its burn at 1.5?

You only need a TWR greater than 1 to overcome gravity at liftoff or when landing. In orbit or any transfer burns you can basically have any lower TWR you like as long as you can cope with the increased burn times. A TWR greater than 1 on a transfer stage usually means you waste a lot of headroom you could have used for more fuel.

[MoarBoostersRUS]

13 minutes ago, Harry Rhodan said:

You only need a TWR greater than 1 to overcome gravity at liftoff or when landing. In orbit or any transfer burns you can basically have any lower TWR you like as long as you can cope with the increased burn times. A TWR greater than 1 on a transfer stage usually means you waste a lot of headroom you could have used for more fuel.

Made this mistake many times when I first started playing. Now most of my probes use ion engines (good luck getting TWR anywhere near 1 with those)

[Spricigo]

4 hours ago, The Flying Kerbal said:

So my questions are, how do you guys figure out what TWR you want your rockets to deliver

If you mean some 'target TWR' I don't have it. Instead I adjust the gravity turn, more powerful rockets start the turn earlier.

Also I don't evaluate efficiency by deltaV to orbit but rather l consider how much funds and time cost to put the payload in orbit

 

[Rocket In My Pocket]

Best launch TWR?

As high as I can get it without exploding.

I'm a little bit impatient. I need to be in space like yesterday people! (Also gravity losses > aero losses.)

[wumpus]

A lot depends on if you are using solid boosters for your first stage.  I've noticed that having TWR>2 (at launch, it will climb and you can't reduce it with SRBs) seems to be a bit excessive, but with SRBs you can get plenty more TWR than liquids.  I might keep adding SRBs until I hit TWR>2, then pull some off until it is back under.

I followed the "cheap and cheerful" challenges (an attempt to get high KEO cargo/fund without recovery), I was under the impression that they were pretty much stuck to ~1.2TWR, but one of Slashy's better attempts appears to be at 1.68 for KSP version 1.0.5.

I'd link to the latter (1.1.3) challenge, but the files don't appear to be maintained and I can't tell what the launch TWR were.  One thing that I am fairly convinced is that you should [almost] never reduce your thrust during initial launch (burn hard to main engine cutoff).  If you have "too much thrust", then you should look into using lighter engines/less SRBs.

My favorite rule of thumb for efficient rocket design (which comes up elsewhere, but not often enough in Kerbal circles) is to at least start with equal amounts of delta-v per stage (this might get altered a bit as kickers provide extra cheap delta-v for a first stage and vacuum engines provide efficient delta-v thanks to tremendous Isp, but it is a great place to start.

 

[bewing]

Wumpus is right. Reducing your thrust almost always reduces your dV efficiency.

[OhioBob]

Low TWR is best for cost efficiency, but I think anything below 1.2 is too low.  I generally aim for something between 1.2 and 1.5 at liftoff, but as close to the middle (1.35) as possible.  Lower on the second stage, probably more like 1.0-1.2.

And although it hasn't been mentioned, I generally like to distribute the propellant about 2:1 between the first and second stages.  I find that if the second stage has a propellant mass approximately equal to the mass of the payload, the first stage has a propellant mass twice that of the second, and the TWR falls within the target ranges, then I get a launch vehicle that performs pretty well.  I use these guidelines for all my launch vehicles.  (These standards are for the stock game and do not apply to larger scales.)

[The Flying Kerbal]

Thanks for all the advice guys... so it seems there really isn't one magic value and just try to find one that works with my flying technique?  I'll be trying a few experimental launches using some of the parameters you use and see what happens. 

[Snark]

12 hours ago, The Flying Kerbal said:

Thanks for all the advice guys... so it seems there really isn't one magic value and just try to find one that works with my flying technique?  I'll be trying a few experimental launches using some of the parameters you use and see what happens. 

Picking the "right" TWR will depend on the characteristics of your ship.  Here are the things you need to balance:

• Gravity losses.  This is a dV waste that your ship incurs because it's spending fuel fighting gravity rather than making the ship go faster.  (Best way to picture this:  imagine if your TWR on the pad were 1.00001.  You'd sloooooooowly lift off, basically just hovering.  Wasting scads and scads of fuel, without actually going anywhere or accelerating the rocket.)
• Aerodynamic losses.  The faster you go, the more dV you lose to drag.  Going too fast too low in the atmosphere means higher aero losses.
• Dead weight.  Engines are heavy.  More-powerful engines are generally heavier.  If you have an insanely high TWR (i.e. "more than you need"), it means you're lugging around lots of dead weight that isn't required, which is a waste of fuel.

Having a higher TWR reduces your gravity losses, but increases aero loss.  Higher TWR will also, generally, speaking, increase your dead weight due to engine mass.

So, what to do?

Important thing to bear in mind:  in general, gravity losses are much higher than aero losses for most "reasonable" ships in KSP.  So you generally want a high launchpad TWR; don't keep it low just to try to avoid aero losses.  The best way to avoid aero loss is to streamline your rocket, not to go easy on the throttle.

A useful bit of math:  during vertical ascent, the most fuel-efficient profile is when the rocket is always climbing exactly at terminal velocity.  If you're going slower than terminal velocity, you'd do better to speed up.  If you're going faster than terminal velocity, you'd do better to slow down.  However, bear in mind that terminal velocity increases, a lot, as you climb because the air thins out.  So for launches on Kerbin, a typical ship never catches up to its own terminal velocity, which means faster is better and high TWR is better.

Climbing at terminal velocity equates to a TWR of 2.  So, a TWR of around 2 on the pad is pretty good.  Note that once your trajectory has curved to be more horizontal than vertical, though, you don't need such a high TWR, because more of your engine thrust is spent in accelerating horizontally (which is good) rather than vertically (which is bad, because vertical climb = gravity loss = bad).  So upper stages are fine for lower TWRs.

I typically design my own ships to have a TWR of about 2 on the pad; upon launch, I immediately crank them over to about 15-20 degrees from the vertical, right off the pad, then my gravity turn just follows all the way up (TWR dropping with each successive stage, as the rocket gets farther from the vertical).  Works great. 

[Tyko]

It seems like I have more success with higher 1.5 - 2 TWR for small rockets and 1.3 - 1.5 TWR for larger rockets. Keep me honest here, but with similar aero designs a smaller rocket will experience more aero drag than a big rocket because surface area doesn't rise as fast as volume.

[OhioBob]

The problem here is that "best" has different definitions, and everybody has their own opinion.  If you gave a concise and unambiguous definition of "best", then the answers would likely start to converge toward a consensus.

For instance, if best means you want to minimize losses and get to orbit using the least amount of delta-v, then I agree with @Snark.  But that's not how I typically define best.  I usually play career mode, so I'm concerned with cost.  I want to put the most payload in orbit for the least funds.  If I'm designing for the best cost efficiency, then my recommendations would be very different than Snark's.

 

[Spricigo]

Yet another issue with suggesting a 'best' TWR is the inability to cover all possibilities of how that TWR will evolve.

SRBs on light craft? TWR raise superfaster. Dropped the boosters? TWR decrease instantly. Large craft with heavy payload? TWR will increase slowly. The list goes on, there is just so many variable to consider it all.

[Lukaszenko]

I find that if you want the most delta-v efficient vehicle, a LOT of TWR is good, with the limiting factor generally being 1) can you fit enough engines 2) can the craft handle the heating. Gravity losses are generally much greater than aero losses. 

My best EVE ascent vehicle, for example, has something between 2/3:1 TWR on the first stage, if I remember correctly. Any less and it wouldn't perform as well, despite the large dead-weight reduction and theoretically higher delta-v. 

 

[wumpus]

On 10/23/2017 at 11:48 AM, Snark said:

I typically design my own ships to have a TWR of about 2 on the pad; upon launch, I immediately crank them over to about 15-20 degrees from the vertical, right off the pad, then my gravity turn just follows all the way up (TWR dropping with each successive stage, as the rocket gets farther from the vertical).  Works great. 

Controlling such a hot launch is a bit different from a more mild liquid-based launch with a TWR around 1.2.  Snark isn't  kidding about "immediately" and for small craft once launch clamps are available you should consider launching them at an angle (although not 15-20 degrees as the starting offset changes things a lot, try 5-10 or even 0-5).  Once the craft hits the transonic region you tend to lose all control of the rocket (although this is great in stabilizing other wise unstable rockets).

* reality check: I've seen at least on high-powered model rocket manufacturer claim that a TWR>5 was needed to stabilizing model rockets using their engines (obviously for rockets with absolutely no guidance or steering controls).  Apparently this "lock in" is a real thing (planes designed for subsonic operation become uncontrollable in transonic speeds: this was a problem for US WWII plane designers, especially Kelly Johnson and the Lightning).

Edited October 24, 2017 by wumpus
s/is/isn't Ouch!

[OhioBob]

On ‎10‎/‎23‎/‎2017 at 11:48 AM, Snark said:

So you generally want a high launchpad TWR; don't keep it low just to try to avoid aero losses.

Although I'm an advocate of lower TWR than Snark, I agree with him on the above.  My choice of low TWR is because I think it is more cost efficient, not because I am trying avoid aero losses.  As long as the rocket is reasonably streamlined, I never fret about aero losses.  And I don't advocate throttling back or thrust limiting a liquid fueled engine just to get within some arbitrary TWR target range.  The reason for using lower TWR is because we can use a cheaper and less massive engine.  But if circumstance force the use of a large engine, i.e. there are no smaller options that will work, then use what you've got and don't throttle back.  Throttling back a large engine means that we're spending big money and incurring large gravity losses to boot.  It's the worst possible scenario.
 

On ‎10‎/‎23‎/‎2017 at 11:48 AM, Snark said:

upon launch, I immediately crank them over to about 15-20 degrees from the vertical, right off the pad

That's really important with high TWR.

There's one number we can look at that's a very good indicator of how efficient our launch was.  That's the amount of delta-v it takes to circularize the orbit when reaching apoapsis.  If you require less than 100 m/s to circularize, then that was a pretty good launch.  On the other hand, if you need, say, 500 m/s to circularize, then you could have done much better.  A high circularization delta-v means that you spent too much time lofting the vehicle vertically.  A trajectory that has a very pronounced arc to it is typically not very efficient.  You want to try to flatten out the trajectory and spend more time thrusting horizontally.  The best launch trajectories are those that end with the trajectory line wrapping all the way around the planet (or close to it) with the apoapsis far downrange of the burnout point.

With a high TWR it is more important as ever to try to flatten out the trajectory as soon a possible.  If you don't, you'll hit the target apoapsis altitude long before you've gain adequate horizontal velocity.  The result will be a highly lofted trajectory requiring a large circularization burn at apoapsis.  So while you may have reduced gravity losses, you've squandered away some of that savings by flying a poor trajectory.

I also think that high TWR is much less forgiving if you do mess up the turn.  If you don't turn soon enough, then the greater velocity that you get with a high TWR just makes it that much harder to execute the sharp turns needed to correct.
 

Edited October 24, 2017 by OhioBob

[regex]

No matter the TWR on the pad (to a point) there's a velocity where you can tip the rocket by 5 degrees, slap the "prograde" button next to the navball, and pretty much get to orbit. I try to build my rockets so that velocity is around 60~65m/s in stock, which is usually around 1.25 TWR or so. Of course, that heavily relies on what the upper stage looks like, how many boosters you're dropping, etc... I am almost always adjusting the trajectory of the upper stage (usually some radial burning to manage apoapsis) and if I can get a circularization burn below 100m/s I call that a win. Never, ever touch the throttle; you can go full-blast the entire launch and that's what most real world rockets do. If you need to throttle it's best to shut off an engine like Saturn V did or choose engines better suited to the task.

[Ultimate Steve]

As far as gravity losses go, your ideal TWR is infinite as to eliminate gravity losses. Low TWR = more gravity losses = less Delta-V available.

Unfortunately, due to the laws of physics, engines require this pesky thing called matter, meaning higher TWR = more engines = more dry mass = less Delta-V.

Now, these two lines intersect somewhere, there exists an ideal compromise where the delta-V wasted on more engines is equal to the Delta-V wasted by hovering in a gravity well for too long.

 

Unfortunately this varies by rocket and there is no one true correct answer.

For general rules of thumb, if your TWR is under 1.1-ish, add more engines (if its a lander you should have a TWR of 2 or more). If it is more than 3, remove engines (unless you want to completely disregard efficiency for POWER!!!).

But, once you are in orbit and won't operate on a collision course with any body, the ideal TWR is 0.0000000000000...1 because the more TWR=more engines=more dry mass=less delta-V and TWR does not matter if you don't need it. At this point it becomes a tradeoff of sitting in front of a computer for four hours doing an ion burn vs. having less Delta-V.

And, let's be honest, nobody likes long ion burns.

At least we don't have *shudder* realistic ion engines...

[Snark]

3 hours ago, wumpus said:

Controlling such a hot launch is a bit different from a more mild liquid-based launch with a TWR around 1.2.  Snark isn't  kidding about "immediately"

Yep. 

Though interestingly, and perhaps somewhat counter-intuitively... I actually find it easier to control the launch (i.e. nail the gravity turn) for the high-TWR craft, than at lower TWR like 1.5 as I used to do.

Reason for this:  For a low-TWR craft, the initial turn is tiny, just a degree or two.  So it has to be nailed just so, because an error of a single degree can make an enormous difference in what the overall trajectory looks like-- it's a really thin margin between "too steep" and "too shallow".

On the other hand... if the TWR is up around 2 and the pitch-right-over-at-launch angle is in the neighborhood of 10-20 degrees ... in that case, a one-degree error is a much smaller percentage, so a small mistake doesn't amplify as much.  So it pretty quickly becomes easy (for me, anyway) to nail the gravity turn pretty well on each launch, even if it's a brand-new ship that I haven't flown before, because all craft at the same rough size and TWR are going to have very similar gravity curves.

On 10/23/2017 at 10:05 AM, OhioBob said:

For instance, if best means you want to minimize losses and get to orbit using the least amount of delta-v, then I agree with @Snark.  But that's not how I typically define best.  I usually play career mode, so I'm concerned with cost.  I want to put the most payload in orbit for the least funds.  If I'm designing for the best cost efficiency, then my recommendations would be very different than Snark's.

An excellent point.  But, as with pretty much everything in KSP, it really depends on play style and design techniques.

For example, yes, I'm trying to use the least dV ... but I'm doing so precisely because I want to save cost.

My experience in KSP has been that "how much dV do I have" is overwhelmingly the single most important limitation of a rocket.  If I want to do a certain mission-- whether it's LKO, Mun landing, trip to Jool, whatever-- there's a certain dV budget that I have to hit.

And dV is hard, due to the tyrrany of the rocket equation.  For a payload of given mass, the required mass (and therefore cost) of the ship goes up exponentially as one needs more dV.  And there never seems to be enough dV to go 'round.

So... if I can save a few hundred m/s off the necessary dV, and by doing so I can get by with a rocket that's 15% lighter... that saves a chunk off the cost.

(Perhaps my perception is skewed by the fact that I tend to rely very heavily on SRBs to get off the pad, which are pretty cheap for the amount of thrust that they generate.  My LFO engines generally kick in once I'm past that brutal initial climb, so I don't need so many of them and they're not an expense.  It's pretty easy and cheap to boost the launchpad TWR of a rocket quite a bit by strapping on some Thumpers.)

[Spricigo]

3 hours ago, Snark said:

For a low-TWR craft, the initial turn is tiny, just a degree or two.  So it has to be nailed just so, because an error of a single degree can make an enormous difference in what the overall trajectory looks like-- it's a really thin margin between "too steep" and "too shallow".r it's LKO, Mun landing, trip to Jool, whatever-- there's a certain dV budget that I have to hit.

That's is why I design mine to be tilted in the launchpad. Then at the appropriate velocity* I just need to change SAS to hold prograde (excerpt a few that are launched Osumi style).

 

*easier to watch than time.

[OhioBob]

8 hours ago, Snark said:

(Perhaps my perception is skewed by the fact that I tend to rely very heavily on SRBs to get off the pad, which are pretty cheap for the amount of thrust that they generate.  My LFO engines generally kick in once I'm past that brutal initial climb, so I don't need so many of them and they're not an expense.  It's pretty easy and cheap to boost the launchpad TWR of a rocket quite a bit by strapping on some Thumpers.)

Your extensive use of SRBs is likely why we differ so much in TWR.  Other than for small rockets, I usually don't use SRBs unless it is to augment my main liquid fueled stage when I need a little extra thrust at liftoff.  My philosophy has always been to trade engine for fuel.  Reducing engine and adding fuel, (1) lowers TWR, (2) increases gravity losses, (3) reduces drag losses, (4) adds delta-v, and (5) reduces cost.  So while my losses are greater, I have the extra delta-v to compensate for it, and overall I've reduced my launch cost.

 

Edited October 25, 2017 by OhioBob

[Helmetman]

In a TSTO recoverable rocket design you usually have lower TWR on the 1st stage. Not always but atleast in most of my own designs since I limit engines on those designs for the lowest dry mass for me to spare dv for landing with the 1st stage.

It needs to land, carry more, have more systems (for i.e.landing)

This is the only rocket design where I go as low as 1.19 twr. But I would never want less then that. Any other rocket design is 1.5 to 2+

Sometimes I build things purely for performance in Sandbox usually. And a high aerodynamic rocket with high twr with a very early gravity turn are not uncommon designs. This works especially well in terms of dv efficiency for SSTO rockets. Although it's iften tricky to get the thrust high enough with the available nodes, esspecially if the rocket in said design is huge.

[Starchaser]

On 10/24/2017 at 3:39 PM, regex said:

Never, ever touch the throttle; you can go full-blast the entire launch and that's what most real world rockets do. If you need to throttle it's best to shut off an engine like Saturn V did or choose engines better suited to the task.

I'm sorry, but I had to take exception to this. I know you said 'most' real world rockets, but with 135 total launches of the 6 orbital shuttles, I think that's a fairly large slice of the total rocketry pie.

The last command issued to Challenger was 'go at throttle up' or, increase to full throttle. This occurred at 2 minutes into it's launch attempt. Increasing to full throttle implies it wasn't at full throttle beforehand. Investigating a little further,  this is in the shuttle's wikipedia entry:

Around 30 seconds into ascent, the SSMEs were throttled down—usually to 72%, though this varied—to reduce the maximum aerodynamic forces acting on the Shuttle at a point called Max Q. Additionally, the propellant grain design of the SRBs caused their thrust to drop by about 30% by 50 seconds into ascent. Once the Orbiter's guidance verified that Max Q would be within Shuttle structural limits, the main engines were throttled back up to 104.5%; this throttling down and back up was called the "thrust bucket". To maximize performance, the throttle level and timing of the thrust bucket was shaped to bring the Shuttle as close to aerodynamic limits as possible.^[78]

 

I know Wikipedia is not considered a great source, but the footnote 78 points to a PDF from NASA regarding the shuttle's wings and the aerodynamic challenges they generated. 

https://www.nasa.gov/centers/johnson/pdf/584730main_Wings-ch4d-pgs226-241.pdf

The throttle-down is discussed on page '233' of the report. (it's an excerpt, the PDF is only 16 pages long.

My apologies if anyone finds this overly pedantic. My intention was to generate interest and conversation. However I am aware that my attempts to do so often come across as pedantic. It's not my intention.