Tweaked aerodynamics ?

[Vindelle_Sunveam]

Hi, with the .24, I feel like my rockets are more likely to feel the burden of atmospheric pressure and tumble out of control when having an eccessive angle of attach relative to the velocity vector. Anyone experienced this feeling ?

[SkyRender]

No, but I have noticed a strange phenomenon wherein the time of day of the launch somehow messes with rocket launch efficiency. A rocket launched at exactly midday will somehow be more efficient going into orbit than one launched at dawn or sunset. This... makes no sense, but my tests have been fairly conclusive. Exact same technique for each launch, but the sunset and sunrise rockets need an extra ~300dV to reach orbit than the midday launches no matter how well I fly them.

[Redrobin]

No, but I have noticed a strange phenomenon wherein the time of day of the launch somehow messes with rocket launch efficiency. A rocket launched at exactly midday will somehow be more efficient going into orbit than one launched at dawn or sunset. This... makes no sense, but my tests have been fairly conclusive. Exact same technique for each launch, but the sunset and sunrise rockets need an extra ~300dV to reach orbit than the midday launches no matter how well I fly them.

What about night launches?

[Pecan]

Err, it's because of heat-expansion, causing, ummm ... nope, can't do it, imagination failure. You're the novelist, you invent an excuse, lol :-)

That really is strange - I shall have to test some time.

[Martijn404]

I found that, on vertical launch, side-boosters that previously dropped nicely straight down now tumble top-inwards thereby destroying the rest of the rocket.

This is the exact same rocket that launched fine yesterday. Same mechjeb settings for maximum reproducability.

[Tracer13]

The photons lubricate the atmosphere and create a Teflon-like effect that helps the rocket achieve a lower drag coefficient. Also the stream of neutrinos from the sun cause an antigravity lifting effect that helps a little.

[Vindelle_Sunveam]

@SkyRender

I'd say it is your rocket fighting against the prismatic effect of rogue photons escaping the atmoshere at very shallow index of refraction (ie. Expanding the space because of... Err... Relalivity), you then need more deltaV to cross more space while drag stays the same at dawn and dusk

@Martijn : I saw that. I had a mainsail destroyed because of that... Did not revert, just had to use the insertion stage earlier

Edited July 25, 2014 by Vindelle_Sunveam

[Felsmak]

That's because you get more help from the Sun's gravity, of course!

[Bloodlance]

This is the problem at hand :

http://forum.kerbalspaceprogram.com/threads/88026-Decouplers-failing-to-apply-ejection-forces-Still%21-%28Post-0-24-1%29

[Pecan]

Here we go again everyone - 0.24.2 is up.

[Bloodlance]

I found a "fix" for it... :D:D:D:D

But yea, its bugged.

Fix in the vid and yes its 0.24.2

sniff

[SkyRender]

I think the "increased delta-V requirement at sunrise and sunset" problem is actually more of an issue with the atmospheric modeling in the game. It only is a problem with stock aerodynamics; FAR launches work properly no matter the time of day (as long as the rocket is aerodynamic to begin with, granted). It's not a new problem, either; I've noticed this effect quite a bit in past releases. I design my rockets to very specific standards (with Kerbal Engineer Redux's help), meaning that I know when something is off in the lift-off part of the program if one of my rockets can't quite reach orbit on its boosters.

I saw dramatically just how true this problem is with stock aerodynamics just yesterday: two identical rockets launched within 3 hours of one another, one at midday and one at sunset. The midday one reached orbit with a little to spare in the boost stage, as per design. The sunset rocket was about 300m/s short of orbital speed when those very same boosters ran out. Their launch profiles were as identical as you can get without resorting to an autopilot, too. I re-ran that sunset launch 3 times just to be sure, but the end result was identical each time: around 300m/s short of orbit when the booster stage burned out. As I said, I've seen it happen in past versions too, but this was the first time I had any solid evidence of the effect.

[Renegrade]

I think the "increased delta-V requirement at sunrise and sunset" problem is actually more of an issue with the atmospheric modeling in the game.

I've launched three test rockets, one at midday, one in the evening and another in the morning.. All three took exactly two of three stages to reach orbit, and the orbits are almost identical. The evening one is actually slightly higher, being a bit higher overall.

Are you um, perhaps using mods that might impact that?

[Vindelle_Sunveam]

Here we go again everyone - 0.24.2 is up.

For tge decoupler thing, I know, I read the .24.2 notes.

Nonetheless, there was no mention of anything that could have affected aerodynamics. The rockets that look like they are tumbling out of control more easily are my usual basic mk1Pod+LF200+lv909+stackDecoupler+LF-800+LVT-45 rocket.

This happens around 25km velocity vector at 20° east, 400m/s and LV45 throttled up at max, ship ponting at 45°.

[cantab]

Wild guess: Maybe the adjustment to Kerbin's day didn't "sync" with the atmosphere. Could be tested with launches from other worlds with atmospheres.

[SkyRender]

I think I need to run a comprehensive battery of tests to be sure of this, but it's something I've observed in the past: sometimes a rocket will just plain underperform compared to how you would expect it to. Granted, a slight difference in launch profile can add or subtract a fair bit of delta-V from a launch, but generally not on a scale that large and definitely not with a launch profile that's utterly identical. Time of day might have been a coincidence in all of this; it could just be that something is getting loaded that gets "stuck" somehow. The discrepancy is too much for it to be mere chance, however...

[SkyRender]

So I figured, given my strange troubles, I would see if I can diagnose this problem as scientifically as possible. And so I ran this experiment.

Test control conditions:

* New Sandbox file, immediately went to VAB to build the rocket

* Exact same craft is used in all tests

* Exact same launch profile is used in all tests

Craft Overview:

1.37 initial TWR

4,876 dV

Structure:

MK16-XL Parachute

Mk1 Lander Can

TR-18A Stack Decoupler

FL-T200 Fuel Tank

FL-T400 Fuel Tank

FL-T800 Fuel Tank

LV-T45 Liquid Fuel Engine

4x TT18-A Launch Stability Enhancers

Launch Profile:

* Full throttle at start

* Stay vertical for first 1000m

* Slowly pitch towards 45 degrees east, reaching the 45-degree mark at 10,000m

* Remain at 45 degrees east until time to AP is 1m5s

* Lower speed to keep time to AP between 1m5s and 1m10s (as close to 1m5s as possible), and follow prograde

* Stop when apoapsis reaches 70,000m to 70,100m (target: 70,050m)

* Circularize at 70,000m to 70,500m (target: 70,050m AP/70,050m PE)

Test Series: Rapid-succession identical launches

Overview: Launch of 5 identical vessels in quick succession (each one was launched ~15 minutes apart).

Observed Data: Fuel levels upon attaining a 70KM orbit. (If failed to reach orbit, indicate with DNRO and list velocity closest to apoapsis.) Launches marked with * are ones where the program was quit and restarted prior.

Hypothesis: There may or may not be a problem which causes vessels to perform differently depending on how they're loaded at launch.

Results:

1. *DNRO (2161.9m/s)

2. DNRO (2100.1m/s)

3. *DNRO (2036.0m/s)

4. DNRO (1983.7m/s)

5. DNRO (2223.2m/s)

Conclusions: Rocket should have reached orbit with fuel to spare every single time, but did not even once. Furthermore, despite identical launch profiles, the final velocity at ~70,050m for all five flights varied immensely, without an absolute pattern. Likely culprit is something relating directly to when the rocket is loaded onto the launchpad, given these results. Further experiments may be warranted.

Edited July 26, 2014 by SkyRender

[ansaman]

I found that, on vertical launch, side-boosters that previously dropped nicely straight down now tumble top-inwards thereby destroying the rest of the rocket.

I have experienced the very same problem with the big boosters and I have been experimenting to see what I can do to ameliorate the phenomena.

[Master Tao]

For boosters, try putting Sepratrons nozzle up and vertically centered on the outsides. Put them in the same stage as the radial decouplers.

[The Jedi Master]

I've seen the booster tumble too.

[Wanderfound]

For boosters, try putting Sepratrons nozzle up and vertically centered on the outsides. Put them in the same stage as the radial decouplers.

You're missing some creative options if you do it that way. Place your Sepratrons right and you can make the boosters do ballet as they fall...

(think Sepratron mounting angles, opposed rockets and spins)

[Renegrade]

Results:

1. *DNRO (2161.9m/s)

2. DNRO (2100.1m/s)

3. *DNRO (2036.0m/s)

4. DNRO (1983.7m/s)

5. DNRO (2223.2m/s)

That final launch was almost at orbit, according to my orbity script it's 2295.92 m/sec at 70.5km. That being said, a rocket with 4800 dv on the pad should DEFINITELY reach orbit. Unless some engine with terrible atmospheric Isp was used.

What was your ascent profile re: speed vs altitude?

[SkyRender]

That final launch was almost at orbit, according to my orbity script it's 2295.92 m/sec at 70.5km. That being said, a rocket with 4800 dv on the pad should DEFINITELY reach orbit. Unless some engine with terrible atmospheric Isp was used.

What was your ascent profile re: speed vs altitude?

For testing purposes, I didn't perform an absolutely-optimal ascent, focusing instead on one that should have provided consistent results. Definitely should have reached orbit each time, however; the profile is close enough to optimal for that. It's outlined above in terms of angle and altitude, and the rocket used is also outlined (it was a very simple vertical stack, so everything is listed in order), meaning you should be able to reproduce it easily if you want to test for yourself.

[Bloodlance]

Here is my fix to the problem

If i didnt post it here, didnt notice it , here :

"fix"

[KerikBalm]

No, but I have noticed a strange phenomenon wherein the time of day of the launch somehow messes with rocket launch efficiency. A rocket launched at exactly midday will somehow be more efficient going into orbit than one launched at dawn or sunset. This... makes no sense, but my tests have been fairly conclusive. Exact same technique for each launch, but the sunset and sunrise rockets need an extra ~300dV to reach orbit than the midday launches no matter how well I fly them.

Well,

#1) I have not noticed this, and I doubt KSP models this

#2) (also addressed to the posts that follwed yours), there is a real world explanation. When its hot, the air is less dense at the same pressure: http://en.wikipedia.org/wiki/Density_altitude

This does not affect ISP (that is pressure dependant, not density), but it does change drag.

Launching at night, when its cooler, can be expected to result in higher drag losses.

As to if it is very significant (especially IRL, where the terminal velocity is really really high) - I doubt it

if KSP models this (I think it uses pressure as density, not a combination of pressure, MW of the atmospheric gasses, and temperature) - I highly doubt it