Efficient take off

[Madscientist16180]

So far, i managed to get my average Dv needs to orbit Kerbin at around 3600 m/s. 

I saw a Scott Mankey vidéo where he was using some mod that have a display for ascent stats, namely percentage of thrust spent on each component of the ascent (gravity loss, drag, starting, etc.).

From the UGI i reckon it was MechJeb but i cannot find such à feature using the mod.

Is there something i'm missing? Can Kerbal engineering give me those stats? I'm trying à MechJeb less career.

I guess i can judge drag by myself, just gotta minimise the flaming effect would probably mean i'm pushing at a sustainable rate that does not loose to much to drag?

I always see those flame at some point in my ascent, normally starting around 35km and dying out at around 45/50 km.

Too much thrust? Going horizontal too fast? 

 

What does people use to record vidéo of KSP? Showing the entire process and how i do it would i guess make it so much easier for the pros to help.

As ever, thank in advance for using your time to help the newbies!

Edited March 15, 2016 by Madscientist16180

[Snark]

42 minutes ago, Madscientist16180 said:

I guess i can judge drag by myself, just gotta minimise the flaming effect would probably mean i'm pushing at a sustainable rate that does not loose to much to drag?

I always see those flame at some point in my ascent, normally starting around 35km and dying out at around 45/50 km.

Don't worry about the flames, they don't hurt anything.  Just ignore them.

Of more concern is if you start seeing Mach effects (feathery white trails at low altitude) too soon.

Basically, you want to be going at terminal velocity all the way up for most efficient ascent.  It's an issue on Eve, but is usually not so much of an issue on Kerbin because the terminal speed goes up pretty fast with altitude.  As long as your launchpad TWR isn't too high, you'll be fine.

Suggestion:  Pick a launchpad TWR you like and stick with it.  1.5 is a pretty good number, it's what I use all the time.  You can go a little higher if you like, some like to go a little lower, I find that 1.5 works well for me.

Then launch, and practice your gravity turn.  You want to be tipped at about 45 degrees by the time you reach 10-12 km altitude, going somewhat over Mach 1.  After that, it's just follow-your-nose at max acceleration all the way to space.

So really the only thing you need to practice is how big of an eastward nudge to give your ship its start on the gravity turn, and when to give that nudge.  Best way to do that is to just practice a lot.  The way you tell whether you're doing it right is whether you're around 45 degrees when you're around 10-12 km.

For 1.5 TWR, what I find works well is to give a very small, gentle nudge practically right off the pad, when I'm going around 20 m/s, and then just follow prograde all the rest of the way.  You'll improve with practice.

[Madscientist16180]

Tartigrade guy at the rescue again. Lol

Thanks for the reply. Looks like i got the basics right, but those flame had me thinking i was loosing lots to drag. 

As you said, gotta practice now

Edited March 15, 2016 by Madscientist16180

[Nich]

From what I have seen if you want to get into the 3250s for dv to orbit you have to circularize around 45-55km and do a hohmann transfer up to you target altitude.  At 45km your lossing a lot to drag and at 55km you losing a lot of oberith.  35km and even 25 km is fine too for efficiency because oberth helps so much but I have a lot of trouble with exploding at 2500 m/s.  The whole range seems pretty break even for me. unless I am launching something extra draggy then I go for 55-65km.

Edited March 15, 2016 by Nich

[mrclucks]

I get my lift stages to have at least 3300 m/s at ASL with a twr around 1.2 and then nudge over around 50m/s around 600m which should get me to 10 degrees at 3km and 45ish degrees at 15km.  That being said my lifter stages usually have enough dv to be used as an initial transfer stage as well if they are not too unruly.  

 

Occasionally I'll build a rocket with a twr of at least 2 and tip over to 10 right as I'm taking off, however I am still aiming for 45 degrees at 15km since that is my magic number for me

[Superfluous J]

The mod that you described sounds like GravityTurn. It does automated ascents to orbit and if you launch many times with the same ship it'll modify itself to be more efficient.

Even if you don't want the game to "play it for you" it's worth installing for a while, just to see what an efficient gravity turn looks like.

[fourfa]

GravityTurn was super educational for me.  It even has data readouts to prove its profile works.  And it uses a very simple algorithm, easily flown by hand if you wish

[Warzouz]

On 15/3/2016 at 6:36 PM, Snark said:

...

So really the only thing you need to practice is how big of an eastward nudge to give your ship its start on the gravity turn, and when to give that nudge.  Best way to do that is to just practice a lot.  The way you tell whether you're doing it right is whether you're around 45 degrees when you're around 10-12 km.

...

A agree with you, but I noticed that generally, it's best to cross the 45° around 8km if payload is reasonably well streamlined.

[Val]

On 15/03/2016 at 5:49 PM, Madscientist16180 said:

I saw a Scott Mankey vidéo where he was using some mod that have a display for ascent stats, namely percentage of thrust spent on each component of the ascent (gravity loss, drag, starting, etc.).

The MechJeb tools you were asking about is not a standard window. You have to create it yourself, but it's not hard.

Click the MechJeb button and find the Custom Window Editor while in flight or SPH/VAB. Click New Window, name it appropriately and select whether you want to see it in flight or in editor. Then add fields from the Recorder category. dV expended, Drag, Gravity and Steering losses.

On 15/03/2016 at 5:49 PM, Madscientist16180 said:

I guess i can judge drag by myself, just gotta minimise the flaming effect would probably mean i'm pushing at a sustainable rate that does not loose to much to drag?

In my opinion drag is very rarely an issue for rockets, regarding efficiency or dV losses.

For rockets gravity losses tend to be the biggest of the 2.

Consider these 2 ascents.

First is a very low drag, airbreathing SSTO with an extremely aggressive ascent path. It reaches 1000 m/s at ~4 km altitude, 1400 m/s at ~10 km, and ~2200 m/s at 30 km.

Drag losses are ~900 m/s dV. Total dV spent ~3400 m/s. (Video)

Now a really high drag rocket, where I do my gravity turn too early and end up spending more time in the lower atmosphere than I should.

Drag losses ~260 m/s. Total dV spent ~3600 m/s. (Video)

Still ends up with minimal drag losses. Doing a proper gravity turn that fits your TWR is much more important than drag.

Edited March 17, 2016 by Val
Accidental post before i was done

[Snark]

4 hours ago, Warzouz said:

A agree with you, but I noticed that generally, it's best to cross the 45° around 8km if payload is reasonably well streamlined.

You may be right, though I suppose it depends on your TWR to some extent-- maybe you tend to build yours with a somewhat higher TWR than I do.  My experience is that if I cross 45 degrees much before 10 km, there's a higher likelihood that I'll end up going too fast nearly horizontally when I'm still down in the soup.

Also, there's a question of how close to the edge you want to walk.  As long as you get to 45 by no higher than around 10-12 km, you'll make it to space fine-- maybe not quite the optimum efficiency, but not shamefully wasteful, either.  i.e. slightly steeper than optimum is fine.  On the other hand, if you err too far in the other way-- i.e. a bit too shallow-- then it can quickly snowball and you end up going nearly-horizontal when you're still too low and you end up plowing through atmosphere halfway around the planet:  failed launch.  I prefer to be a bit conservative; I'd rather risk a slightly-less-than-optimum flight from being too steep than a much-less-than-optimum from being too shallow.

[Warzouz]

@Snark. Yes you're right, that depends on how we build our rockets.

That why I like 1.0 aero. It's not as obvious as before. I remember nearly always using Mechjeb before 1.0, not because I didn't know how to go to LKO, but because I found that not very interesting. Since 1.0, I nearly never used MJ to go to LKO (well sometime to do the circularizing burn) because it's much more interesting than before.

[Plusck]

If you play modless (which I would recommend highly for a new player), the main issue with getting your ascent right is (imho) the problem of getting all the necessary info on the screen at the same time. The easy answer, of course, is to get a mod like KER, but then you ruin that whole process of getting a feel for what your rockets can do, and the feeling of success getting down and back from the Mun, intercepting other planets the first time and so on.

From my experience, one of the most crucial bits of info is the "time to apoapsis", which you can only see by mousing over the screen on the map view. However, you also need to watch your staging, so you absolutely have to return to external view all the time. A mini-map on the external view would be a nice addition to stock (for this and for rendezvous, it would be priceless, but I digress...). 

All of my most efficient launches have had a "time to apoapsis" which averaged about 30-35 seconds over the launch.

Obviously it starts at zero on launch, and should rise to somewhere around the 30s mark as you approach 45° and continue rising until you lose your first stage. Since the middle/upper stage tends to have a lower TWR, it'll start to drop again as you start to head towards the horizon. If it drops below about 20s, you're probably not going to make it without pointing up more and away from prograde, which needs significant control authority to prevent flipping.

This also gives an early indication that you've messed up and headed too steeply up: if you have a minute to apoapsis at any time except at the very end of the ascent, you haven't been agressive enough turning off the launchpad.

[OhioBob]

Liftoff TWR is a topic that has been much debated.  I've heard people say they use numbers ranging from about 1.2 to 2.0.  Generally speaking, high TWR will make your rockets more Δv efficient, while low TWR will make your rockets more cost efficient.

The advantage of high TWR is that it reduces gravity losses.  While sitting on the launch pad, it takes a TWR of 1 just to counteract the force of gravity.  If you have a TWR of 2, then 50% of your thrust is used to cancel out gravity.  However, if your TWR is 1.2, then 83% of your thrust is used to cancel out gravity.  With a high TWR, far more of your thrust is used in gaining velocity than when you have a low TWR.  One of the disadvantages of high TWR is that you'll be moving faster in the low atmosphere, therefore you'll have higher drag losses.  However, drag losses are less than gravity losses, so it is a good trade to accept higher drag losses in exchange for lower gravity losses.  With a high TWR it might be possible to get to orbit using only 3200 m/s Δv, maybe even less.

The biggest disadvantage of high TWR is that a rocket's most expensive part is its engine.  To get a high TWR a rocket requires a large and expensive engine.  Your cost per ton of payload delivered to orbit is generally least when using as small an engine as possible, i.e. a low TWR.  To see this, let's consider an example.  Let's say we have a 100 ton launch vehicle powered by a Twin-Boar engine.  This rocket will have a liftoff TWR of about 1.9.  Suppose it can deliver a 23-ton payload (a payload fraction of 0.23) using 3200 m/s Δv.  Those numbers sound pretty good, right?.  Let's say we now start packing on more fuel tanks while keeping the same engine.  Suppose we bring the total mass up to 136 tons, which lowers the liftoff TWR to 1.4.  Because of the lower TWR, we'll experience greater gravity losses.  Let's say it now takes 3400 m/s Δv to reach orbit.  However, since we've packed on all that extra propellant, we can lift a larger payload.  Suppose we can now deliver a 29-ton payload (a payload fraction of 0.21).  The low TWR rocket is less efficient in terms of both Δv and payload fraction.  However, since we gained the extra 6 tons of payload at the cost of adding only relatively cheap fuel tanks, our cost per ton of payload is actually less.  The moral of the story is, be careful about what metrics you look at when determining what the most efficient rocket is.  If your goal is cost efficiency, then don't be wooed by things like low Δv.

The example above uses made up numbers but, from my experience, they're about what can be expected.  I've found that maximum cost efficiency is generally gained, in theory, with a liftoff TWR of about 1.2.  However, I find that rockets with such a low TWR require so many fuel tanks to be stacked on top of one another that the rocket becomes very slender, wobbly, and difficult to steer.  I generally target a TWR in the 1.3-1.5 range, which I think is a good compromise.  For the second stage I generally try to have an initial TWR of about 1.0-1.3.  These are just my personal guidelines (other people are bound to disagree), but I've found them to work rather well.

Using my TWR guidelines, I generally like to be pitched over about 45^o at 10-12 km, and nearly horizontal by about 35 km.  One metric I use to determine the efficiency of my ascent is the magnitude of my apoapsis circularization burn.  If I have to make a large burn, say >200 m/s, then my ascent was too highly lofted.  My trajectory reached the target apoapsis before I gained enough horizontal velocity.  I'm happy anytime my apoapsis burn is not more than about 100 m/s.

Something else to be aware of is that the Δv require to reach orbit is generally greater when using small rocket parts.  With 2.5 m parts and a streamlined payload, I'm almost always right around 3400 m/s.  However, when building a small rocket using 1.25 m parts, I generally figure about 3600 m/s.  Also, early in a career game before the discovery of fairings, payloads may not be very streamlined.  When launching a draggy payload on a small rocket, I generally figure about 3800 m/s. 

 

Edited March 19, 2016 by OhioBob

[wumpus]

On 3/15/2016 at 1:36 PM, Snark said:

Basically, you want to be going at terminal velocity all the way up for most efficient ascent.  It's an issue on Eve, but is usually not so much of an issue on Kerbin because the terminal speed goes up pretty fast with altitude.  As long as your launchpad TWR isn't too high, you'll be fine.

Is this still true?  I was under the impression that flying at terminal velocity was an artifact of the <1.0 aero model (certainly related bit where you wanted a TWR=2.0 at all altitudes is not).  I had to check the date to make sure I wasn't reading a necro thread.

5 hours ago, Plusck said:

If you play modless (which I would recommend highly for a new player), the main issue with getting your ascent right is (imho) the problem of getting all the necessary info on the screen at the same time. The easy answer, of course, is to get a mod like KER, but then you ruin that whole process of getting a feel for what your rockets can do, and the feeling of success getting down and back from the Mun, intercepting other planets the first time and so on.

While many players may have had fun learning on stock KSP, I don't think the "more boosters" approach is an option with KSP 1.0.5 and beyond.  Rockets are fickle, flimsy, and like to flip over.  You want rockets with the right TWR and the right delta-v, and you either need to calculate it yourself (an annoying physics homework problem) or let KER do it for you.  I'd recommend KER.

Edited March 17, 2016 by wumpus
missed the TWR=2.0 for KSP<1.0

[OhioBob]

10 minutes ago, wumpus said:

Is this still true?  I was under the impression that flying at terminal velocity was an artifact of the <1.0 aero model (certainly related bit where you wanted a TWR at all altitudes is not).  I had to check the date to make sure I wasn't reading a necro thread.

You still want to fly at less than or equal to terminal velocity (the physics for why that is true hasn't changed); however, since 1.0, reaching terminal velocity while flying a nominal ascent from Kerbin is virtually impossible, making it a moot point.  However, it can still be an issue when ascending from Eve, though not nearly as big a problem as it use to be.  At times of high acceleration, when nearing stage burnout, it's still possible to exceed terminal velocity on Eve.  

[Snark]

2 hours ago, wumpus said:

Is this still true?  I was under the impression that flying at terminal velocity was an artifact of the <1.0 aero model

Yep, still true.

What changed with 1.0 is that terminal velocity is now more complicated.  Used to be that all rockets had exactly the same terminal velocity for a given altitude on a given planet, so you could just memorize a table:  On Kerbin, terminal velocity is 103 m/s at sea level, 110 m/s at 1000m, 121 m/s at 2000m, etc.

Now your terminal velocity depends on your shape, mass, and so forth.  Which means it's a non-trivial quantity anymore-- depends crucially on your ship design.

It's also the case that for a well-designed rocket, terminal velocity is a lot faster now than it was pre-1.0.  Which is why as a practical matter you don't need to worry about exceeding it too much (unless you're on Eve), because unless you have a stupidly high TWR, terminal velocity will increase faster than your rocket's velocity will, so you don't need to worry too much about catching up with (or outrunning) it.

[Madscientist16180]

6 hours ago, Plusck said:

If you play modless (which I would recommend highly for a new player).

Kinda depend on what you mean by new
Been playin since a bit before Christmas, i think i learned quite well, already visited Duna manned, and every other places unmanned (i admit Moho was surprisingly challenging, not getting there, but actually stayin within the SOI lol)

So, might be a new user, i'm more trying to make everything as efficient as possible rather than learn the very basics (that will not stay true when we start talking planes, i really do not get it much lol).
Yet, every single post was very informative.

Also, i grasp concept fairly easily. Background thing i would expect, trained chemist, science lover. I just got into involved physics and this game is awesome to grasp orbital concept.

There's been many answers, i won't quote each one of you to thank you all, but thank you all lol! All your answer kinda showed me that i was doing thing right, just gotta thrust my rocket more and stop manually managing the ascent to see how it goes now
Without further ado, i'll start the game and try a few thing! The goal for tonight, an efficient natural gravity turn, let see how it goes!

On 2016-03-15 at 2:49 PM, Nich said:

From what I have seen if you want to get into the 3250s for dv to orbit you have to circularize around 45-55km and do a hohmann transfer up to you target altitude.  At 45km your lossing a lot to drag and at 55km you losing a lot of oberith.  35km and even 25 km is fine too for efficiency because oberth helps so much but I have a lot of trouble with exploding at 2500 m/s.  The whole range seems pretty break even for me. unless I am launching something extra draggy then I go for 55-65km.

Circularizing inside the atmosphere? I would never had thought about trying this...
I will, if only for the fun of it! Exploding rocket always makes my son laugh lol.

[Superfluous J]

34 minutes ago, Madscientist16180 said:

Circularizing inside the atmosphere? I would never had thought about trying this...

I will, if only for the fun of it! Exploding rocket always makes my son laugh lol.

While more efficient, unless I've got some sort of autopilot I don't circularize within the atmosphere for one specific reason: You can't time warp. Sure, you can 4x time warp but when waiting for a ship to get to the other side of the planet, I want a bit more than that.

I'll happily bring some extra fuel to avoid that wait.

[Madscientist16180]

14 minutes ago, 5thHorseman said:

While more efficient, unless I've got some sort of autopilot I don't circularize within the atmosphere for one specific reason: You can't time warp. Sure, you can 4x time warp but when waiting for a ship to get to the other side of the planet, I want a bit more than that.

I'll happily bring some extra fuel to avoid that wait.

Quite a good reason, yet i see the challenge of circularizing as low as possible in the atmosphere as pretty cool.
Not something i'll try in my career though lol!

[Plusck]

1 hour ago, Madscientist16180 said:

Kinda depend on what you mean by new
Been playin since a bit before Christmas, i think i learned quite well, already visited Duna manned, and every other places unmanned (i admit Moho was surprisingly challenging, not getting there, but actually stayin within the SOI lol)

So, might be a new user, i'm more trying to make everything as efficient as possible rather than learn the very basics (that will not stay true when we start talking planes, i really do not get it much lol).
Yet, every single post was very informative.

Ha! Well, while re-reading my post before clicking submit, I was wondering whether to add an "-ish" to the "new".

I played the demo to death at several times over the years, tried the latest demo back in October and eventually gave in and bought the game. So I was definitely a new (-ish) player in October last year. Played hundreds of hours per week to start with, watched a lot of videos, agreed with some and not with other things I saw. Which brings me to another point:

5 hours ago, wumpus said:

While many players may have had fun learning on stock KSP, I don't think the "more boosters" approach is an option with KSP 1.0.5 and beyond.  Rockets are fickle, flimsy, and like to flip over.  You want rockets with the right TWR and the right delta-v, and you either need to calculate it yourself (an annoying physics homework problem) or let KER do it for you.  I'd recommend KER.

Admittedly, I cut my teeth on the 0.23 demo (I think it was) and really started playing the real game with 1.0.4. So I did do a lot of "old style" things like forcing the gravity turn at 10km to start with. Still, I only started using KER after completing the stock science tree in 1.0.5, and by that time had sent probes (but not beings) to the first three planets, and even managed a sun-orbit-near-Dres rescue mission without having my dv spelled out for me.

So if you're willing to think about why things don't work all that well (I killed Jeb early with an unrecoverable descent, and decided it was an R&D risk and left him dead...), I really don't think that you need to go straight to KER or MechJeb. Career really is designed to lead you through the parts, what they can do and what you need to plan for when using them, and I think it does it very well (or at least, it did for me).

1 hour ago, Madscientist16180 said:

Circularizing inside the atmosphere? I would never had thought about trying this...

I will, if only for the fun of it! Exploding rocket always makes my son laugh lol.

I had a couple of ships to prepare just after reading Nich's post, so I noted numbers and made a ton of screenshots just to check.

I didn't circularise at 55km, but I did have a very flat trajectory (basically coasting from 50km to 75km, waiting for a 50 m/s circularisation burn). I set the circularisation burn as soon as I cut the engines, then reset it 5 minutes (game time) later after escaping the atmosphere since the numbers necessarily changed. Total losses to drag from that long coast up? 3.5 m/s...

Album for that last attempt (with laughable design for the actual ship) is here: http://imgur.com/a/ksLj8

Edited March 18, 2016 by Plusck

[Nich]

4 hours ago, Plusck said:

 

I had a couple of ships to prepare just after reading Nich's post, so I noted numbers and made a ton of screenshots just to check.

I didn't circularise at 55km, but I did have a very flat trajectory (basically coasting from 50km to 75km, waiting for a 50 m/s circularisation burn). I set the circularisation burn as soon as I cut the engines, then reset it 5 minutes (game time) later after escaping the atmosphere since the numbers necessarily changed. Total losses to drag from that long coast up? 3.5 m/s...

Album for that last attempt (with laughable design for the actual ship) is here: http://imgur.com/a/ksLj8

I never thought of calculating drag like that very interesting.  I knew the drag was small I had no idea it was that small lol

[numerobis]

In my ascents on the cheap-and-cheerful challenge (which I keep meaning to get back to) I find I get about 30 m/s of aerodynamic drag total above 30km, and that's with a low-TWR which requires me to keep the nose above prograde. Drag is really not a problem up there. So much so that I recommend taking off all your nose cones, at least on the upper stage: you're paying more to accelerate their mass than they're saving you in drag.

In an ascent, any thrust you produce that is above the horizon is wasted; only the horizontal speed matters. The only reasons to thrust upwards at all are to avoid hitting the ground, to reduce drag losses by going higher above the atmosphere, and to reduce your peak skin temperature (atmospheric heating falls linearly with density, and grows with the cube of speed, which puts certain limits on your altitude).

[Nich]

Nose cones only cost 1 dv from what I have seen from in KER. They add a lot of stability and anti flipping tendencys. The cost on the other hand is hard to justify