Huge difference in vehicle performance after Reverting To Launch

[n.b.z.]

I've had a longstanding suspicion that the performance of some vehicles is affected - sometimes radically - after reverting to launch.

I have often used "Revert to launch", numerous times, while trying to figure out the ascent profile for a rocket or plane. When I think I'm done and have a reliable profile worked out, I later find that the profile does not work at all after loading the same vessel from the VAB/SPH.

Recently I made a relay satellite, and slapped some launch stack together to get it to space. At the point when the first stage is spent, there is a velocity difference of about 200 m/s between "load from VAB" and "revert to launch"!

What the krak?!

So today I made a very silly single stage rocket just to research this phenomenon. To rule out other factors - such as my manual control inputs - I launched it straight up, with SAS HOLD already activated before liftoff.

Loading this thing from the VAB:

Highest Speed Achieved: 427 m/s
Highest Altitude Achieved: 7108 m

Then, reverting to launch:

Highest Speed Achieved: 836 m/s
Highest Altitude Achieved: 38823 m

While I suck at math, even I can see that the difference between 7108 and 38823 doesn't exactly look like a rounding error.

So I'm now certain that something is borked. However, I have not had the time to research the exact cause.

I am guessing that for some parts, drag is calculated even when these parts are inside a fairing. After reverting, this seems to change and the fairing then works as expected.

I have not made a bug report, as I'm not sure if this is already known. And I would like other fellow Kerbonauts to verify this issue, to ensure that the bug resides in the game and not in my head.

(I am using the medium-sized fairing. I did not see the same effect when I tried this with the small one.)

Short Imgur album showing the numbers, and the silly rocket:
http://imgur.com/a/l2ZaN

 

[magnemoe]

Tested this and got 868 m/s and 41.9 km on first launch in new game. 
opened fairing before launch and got 362 m/s and 6148 meter. 
Looks like you was right in that fairing was disabled somehow. 

[Nich]

1.2 or 1.1.3?

[Kerbal Kommander]

Thus is something that should be reported in the bug tracker.

[n.b.z.]

1 hour ago, Nich said:

1.2 or 1.1.3?

My test was done in 1.2 preview, but I've had this "suspicion" for a longer time.
 

1 hour ago, Kerbal Kommander said:

Thus is something that should be reported in the bug tracker.

I agree, but would like to see this confirmed by other players first. So I'd be grateful if someone could quickly slap together some test rocket with some very draggy parts inside a medium-size fairing.

The testing itself is easy: hit "Launch" in the VAB, launch rocket straight up, write down max. speed and altitude. Then, revert to launch, repeat same launch procedure, and compare the speed and altitude.

The test rocket shown in my album was specifically designed to react strongly to drag differences:

1. Ridiculously high TWR off the pad (so that drag losses are much more pronounced than gravity loss)
2. An insanely draggy part inside the fairing (largest relay antenna dish pointed upwards, which should be an utter insult to aerodynamics if not placed inside a working fairing)

It also has a short burn duration, so that test runs to maximum altitude don't last forever.

[Pecan]

Ha, my own laziness has saved me from this.  I have the problem that MJ disappears if I revert to launch, coming back when I go to the VAB.  Accordingly I always revert straight to VAB.

[Kerbal Kommander]

I cannot confirm this.  Did several launches with several rockets with and without fairings, and results are very similar for each flight reverting to launch, and loading from VAB.

[n.b.z.]

7 minutes ago, Kerbal Kommander said:

I cannot confirm this.

While this puzzles me, thank you for your effort testing this.

So the bug really is in my head? Oh please, not another one - I'm already bonkers enough!

[Nich]

I did some testing with 1.1.3 and the 2.5 m aero shell preformed identically after revert (within 1%) so this is new for 1.2