Problems with a payload.

[Anquietas314]

You're measuring efficiency in delta-V to orbit. If you measure by payload fraction, lower TWR rockets come out ahead. More dV expended, but less fuel consumed (this results in lower cost, too).

Hey, it's not my model. Go annoy arkie , although I will say this seems implausible given more deltaV fundamentally means more fuel used (for a given rocket), never mind the fact payload mass means much higher dry mass, so your TWR doesn't increase as quickly. In that model, that means much less fuel efficiency. Arkie's model does assume SSTO, so of course you'll have less efficiency loss in practice with multi-stage rockets, but then as we all know SSTO is the cheapest if not the most efficient in KSP, since you can return the entire rocket to KSC for (almost) a full refund .

Edited January 13, 2015 by armagheddonsgw

[Red Iron Crown]

Hey, it's not my model. Go annoy arkie , although I will say this seems implausible given more deltaV fundamentally means more fuel used (for a given rocket), never mind the fact payload mass means much higher dry mass, so your TWR doesn't increase as quickly. In that model, that means much less fuel efficiency.

It seems paradoxical at first glance, but it's not. Lowering the TWR of a given rocket means less engine mass, which means less dry mass, which means more dV for the exact same amount of fuel. Up to a certain point, that greater dV per unit of propellant offsets the greater gravity losses. I have found empirically that an initial TWR of around 1.3 for staged rockets yields the best fuel efficiency even though the dV to orbit goes up by a bit over 150m/s compared to a TWR of 1.6 or so. Cost is reduced even more, because less fuel burned is combined with lower engine costs.

[Jouni]

No, this actually has a large effect on efficiency since if you have low TWR, you're suffering from quite large gravity losses all the way up. Of course the reverse situation is also very inefficient - crazy high TWR (and no throttling to compensate) will make you suffer drag losses.

Efficiency matters, if you're far away from Kerbin. On the launchpad, everything is so cheap that the minor 200-300 m/s losses from a low initial TWR don't matter at all, except in some rare circumstances.

[Anquietas314]

Efficiency matters, if you're far away from Kerbin. On the launchpad, everything is so cheap that the minor 200-300 m/s losses from a low initial TWR don't matter at all, except in some rare circumstances.

It's much more than 300m/s. From the model I linked, that 70% is (roughly) 70% of the total to get to orbit. For kerbin that's about 3,500 m/s, give or take, assuming SSTO as mentioned above. Granted a lot of that is unavoidable due to gravity turn/atmosphere etc, but a decent chunk (roughly a third?) of it is.

[wibou]

HOLY COW, that thing is _massive_. You use SRB to throw used stages backward? THIS THING IS AWESOME! I LOVE IT!!

After watching the video, it is clear the wobbling start appearing when you end up with the large and heavy bottom and the narrower-but-heavy top linked by a weak narrow central coupling... the physic engine dislike this, obviously.

You should be able to (somehow) fix this by adding struts between the central "tail" bottom part and the heavy top part (10:43 in your video). Adding 4 struts at ~60 degree between the top 4 orange tank linked to the start of the kerbodine gray tank should really help I believe... You might have trouble to add those in VAB when your 6 other bottom orange tanks are present, but it would definitely help (although I suspect it won't be perfect).

Another way would be to make sure you use the same type of tank (i.e. don't mix rockomax and kerbodyne) along all the length of the rocket... The wobbling is basically caused because the bottom "mass" is larger than the coupling so using a tank as large as the decoupler should help a lot. However, since you have a very large payload atop, it might still wobble anyhow... you would need to test.

[Jouni]

It's much more than 300m/s. From the model I linked, that 70% is (roughly) 70% of the total to get to orbit. For kerbin that's about 3,500 m/s, give or take, assuming SSTO as mentioned above. Granted a lot of that is unavoidable due to gravity turn/atmosphere etc, but a decent chunk (roughly a third?) of it is.

That model is about horizontal ascents from airless bodies, while we're talking about vertical ascents from bodies with an atmosphere. Stock atmosphere pretty much destroys the benefits from having a high TWR, because terminal velocity is so easy to reach. During the first 10 km of the ascent, rockets with a high initial TWR avoid significant gravity losses, but lose a major fraction of the gains to increased drag. Beyond the first 10 km, even rockets that launched with a low TWR will probably have high enough TWR that the differences to high-TWR rockets become negligible.

[Anquietas314]

That model is about horizontal ascents from airless bodies, while we're talking about vertical ascents from bodies with an atmosphere. Stock atmosphere pretty much destroys the benefits from having a high TWR, because terminal velocity is so easy to reach.

First, horizontal ascents are more efficient than vertical ones due to Oberth effect and a few other details; think of the model as a sort of lower bound on deltaV and upper bound on efficiency losses. As I said earlier, you can directly compare the gravity and drag losses between low and high TWR using MechJeb. I would myself, but alas, gaming machine is broken . In the atmosphere, a "high" TWR is anything over 2. We're talking about a rocket with a TWR significantly less than 2, and you really won't reach terminal velocity (or even near it) at all in that situation (follows pretty much from its definition).

EDIT: I will add that, with FAR (and IRL), a significant part of the reason lower TWR rockets perform almost as well is due to the lifting body effect and starting the gravity turn much earlier than you do in stock. I suspect this may be part of the confusion here.

/EDIT

As for the first 10km, you are aware that pretty much half of the deltaV spent to get to orbit is done so during this part of the ascent, in the optimal case? Low TWR means large inefficiencies due to gravity losses in this stage; in my experience that can be as much as 50-60% of the deltaV spent with a very low TWR rocket. The excessive drag losses can be avoided by throttle control; KER is especially helpful for that, mostly due to the atmospheric efficiency and terminal velocity readouts, which of course will also tell you how much you're losing due to gravity losses.

Edited January 13, 2015 by armagheddonsgw

[Jouni]

First, horizontal ascents are more efficient than vertical ones due to Oberth effect and a few other details; think of the model as a sort of lower bound on deltaV and upper bound on efficiency losses. As I said earlier, you can directly compare the gravity and drag losses between low and high TWR using MechJeb. I would myself, but alas, gaming machine is broken . In the atmosphere, a "high" TWR is anything over 2. We're talking about a rocket with a TWR significantly less than 2, and you really won't reach terminal velocity (or even near it) at all in that situation (follows pretty much from its definition).

Horizontal ascents don't really work on planets with an atmosphere. Due to drag, you usually want to raise the apoapsis above 0 m long before you have reached orbital speed.

Terminal velocity is overrated. A high-TWR rocket (meaning a rocket that can reach 90% of terminal velocity at 10 km) might lose 800 m/s to gravity and 600 m/s to drag before reaching 10 km. If we have a low-TWR rocket that spends 50% longer climbing to 10 km, it might lose 1200 m/s to gravity (the same gravity for 1.5x longer) and 400 m/s to drag (1.5^2 times less drag for 1.5x longer). This 200 m/s difference corresponds quite well to my experiences from designing stock launch vehicles in 0.22 to 0.23.5.

Edit: Some actual numbers using a Skipper and with MechJeb set to limit to terminal velocity.

[table]

[tr]

[td]Initial TWR[/td]

[td]Time to 10 km[/td]

[td]Gravity losses[/td]

[td]Drag losses[/td]

[td]Total losses[/td]

[/tr]

[tr]

[td]1.69[/td]

[td]1:15[/td]

[td]723 m/s[/td]

[td]551 m/s[/td]

[td]1274 m/s[/td]

[/tr]

[tr]

[td]1.25[/td]

[td]1:45[/td]

[td]1015 m/s[/td]

[td]407 m/s[/td]

[td]1422 m/s[/td]

[/tr]

[tr]

[td]1.10[/td]

[td]2:06[/td]

[td]1217 m/s[/td]

[td]356 m/s[/td]

[td]1573 m/s[/td]

[/tr]

[/table]

The difference between a high-TWR rocket and a low-TWR rocket was 150 m/s during the initial 10 km, while the difference to a very low-TWR rocket was 300 m/s.

Edited January 14, 2015 by Jouni

[Anquietas314]

The difference between a high-TWR rocket and a low-TWR rocket was 150 m/s during the initial 10 km, while the difference to a very low-TWR rocket was 300 m/s.

Hmm, okay. What was the payload fraction of that rocket? If you just stuck a fuel tank to a capsule/probe and added the skipper, TWR would increase quickly enough that the difference would indeed be small.

[Jouni]

Hmm, okay. What was the payload fraction of that rocket? If you just stuck a fuel tank to a capsule/probe and added the skipper, TWR would increase quickly enough that the difference would indeed be small.

If you know the initial TWR and the Isp curve, the rest of the ship doesn't matter. There's no difference between unused fuel and payload.

[Anquietas314]

If you know the initial TWR and the Isp curve, the rest of the ship doesn't matter. There's no difference between unused fuel and payload.

Except you're only testing a rocket to 10km, which can easily be a very small rocket that's incapable of going further, so it does matter.

EDIT: The reason it matters is because, with rockets with a larger payload fraction, the TWR will increase much more slowly, which means you'll spend longer far below terminal velocity, meaning gravity losses are much more significant.

This is also ignoring the gravity losses from performing your gravity turn, where drag losses become more and more irrelevant as you ascend. Depending on TWR in this phase, you may or may not be able to perform it efficiently while maintaining an ascent (demonstrated by OP's rocket).

Edited January 14, 2015 by armagheddonsgw

[Jouni]

EDIT: The reason it matters is because, with rockets with a larger payload fraction, the TWR will increase much more slowly, which means you'll spend longer far below terminal velocity, meaning gravity losses are much more significant.

The rate of TWR increase is (almost) a function of initial TWR and engine Isp.

Climbing the first 10 km requires 1500-1800 m/s of delta-v, depending on the initial TWR. With the typical first stage engines used in KSP, the amount of fuel required is around 35-40% of the launch mass. As a result, TWR increases by roughly 50-70%.

This is also ignoring the gravity losses from performing your gravity turn, where drag losses become more and more irrelevant as you ascend. Depending on TWR in this phase, you may or may not be able to perform it efficiently while maintaining an ascent (demonstrated by OP's rocket).

This can be a problem with rockets that stage too frequently or too soon. If the TWR never rises very high, the rocket has to fight gravity all the time. A more reasonably staged rocket builds up climb rate and the time to apoapsis during the high-TWR phase at the end of a stage, allowing the next stage to burn mostly horizontal during the initial low-TWR phase.

Asparagus staging is particularly sensitive to low initial TWR, because it tries to minimize the variation of TWR. Rockets without fuel crossfeed are more forgiving, as their TWR grows more towards the end of a stage.

[Anquietas314]

Climbing the first 10 km requires 1500-1800 m/s of delta-v, depending on the initial TWR. With the typical first stage engines used in KSP, the amount of fuel required is around 35-40% of the launch mass. As a result, TWR increases by roughly 50-70%.

This would be fine, if you did indeed build a rocket that fits this specification; you have so far provided no evidence of that, since your description was "a Skipper and with MechJeb set to limit to terminal velocity", which says nothing of the deltaV capacity of that rocket, what fuel tank was used, etc.

For example, say you had just enough fuel to hit 10km. Using a mk1 command pod as the control part, a single Rockomax X200-8 would be just about enough fuel, and of course you can set the TWR via the thrust limiter. That makes your "payload fraction" at 10km, including engine/tank, is 4.5/9.3 = 48%, or in other words fuel is 52% of your launch mass. That means your TWR increase is approximately 108% - much more than your estimate.

[Jouni]

For example, say you had just enough fuel to hit 10km. Using a mk1 command pod as the control part, a single Rockomax X200-8 would be just about enough fuel, and of course you can set the TWR via the thrust limiter. That makes your "payload fraction" at 10km, including engine/tank, is 4.5/9.3 = 48%, or in other words fuel is 52% of your launch mass. That means your TWR increase is approximately 108% - much more than your estimate.

The Skipper/X200-8/Mk1 combination weights 8.3 tonnes, out of which 4 tonnes (48%) is fuel. When launched vertically with a low TWR, it has around 2200 m/s of delta-v. With a very low initial TWR, the rocket will burn around 3.5 tonnes of fuel (42%) to reach 10 km. This will increase the TWR by around 70%.

[Anquietas314]

The Skipper/X200-8/Mk1 combination weights 8.3 tonnes

Blame the outdated KSP planner I had to use for that. Apparently the skipper was buffed in 0.24. Still, the result there is sensitive to the command pod used (and thus payload fraction). The previous 5 posts could have been avoided if you just posted the specs of the test rocket as asked. Please do that. Even a screenshot will do.

[Jouni]

Blame the outdated KSP planner I had to use for that. Apparently the skipper was buffed in 0.24. Still, the result there is sensitive to the command pod used (and thus payload fraction). The previous 5 posts could have been avoided if you just posted the specs of the test rocket as asked. Please do that. Even a screenshot will do.

As I've said many times, the specs don't matter. If two rockets have the same Isp curve and the same initial TWR, they will burn the same fraction of their mass when climbing to 10 km.