Do perfectly good designs not work the next day?

[Brainlord Mesomorph]

Does it seem to anyone else like a perfectly functional  ship you design one night simply doesn’t work the next day?

This week, I’ve been trying to design a low tech Duna lander, and I’m on my fifth version of various short squat two stage things that will make LKO. (I figure that means it has more than enough aerodynamics,dV and TWR the land on Duna with parachutes, and still make orbit again).

But it seems like five times now I designed something that makes orbit 3 or 4 or 5  times, and decide “yes that will do” and I’m done for the night, and then the next day the damn thing won't make orbit!

Now I have no idea what I’m doing differently, I can’t imagine it’s the game, but I actually thought about writing this post last night before I decided to design version 5! And now version 5 won't make orbit! 

Does this happen to anyone else?

Edited April 6, 2016 by Brainlord Mesomorph

[Empress Neptune]

I've been through something similar a few times. I think I just wasn't thorough enough in testing and I was made to pay for it later. The last time involved a rocket that I thought was stable, but was actually horribly unstable during ascent. During the few tests I did, I happened to fly it just right, somehow avoiding even the tiniest amount of excessive pitch. When I was done testing I exited KSP. Then when I returned to the game I couldn't control the rocket at all. It would tumble whenever I tried to turn it. After many, many attempts I finally managed mimic the earlier flights, but I decided that the original design was too risky to leave unfixed.

 

I think of it as good luck gone bad. Good luck because you happen to find the best case on your first flight, but it goes bad when it makes you think your design is better than it really is.

[Ten Key]

I've had this happen. In particular, I had a rocket with a somewhat marginal staging arrangement that suddenly stopped working after a dozen or so successful launches. My best guess is that it has something to do with the way the game throttles the physics in response to game conditions. If a design is borderline, it may only be working because the game is backing off the calculations a bit. 

[SpacedCowboy]

I think everyone has the T-shirt for this. It's late at night, you've just designed the best space plane EVER!    Next day,,,, meh, not so hot. 

[Kyrt Malthorn]

I do sometimes just have a Bad Launch Day.  Or a Bad Landing Day.  I will sometimes allow done finicky launches to be standardized if they don't carry kerbals....

 

Basically I can'play KSP when I'm exhausted.  (Which is sadly frequent with my work schedule.)

 

On the other hand, I have had weird physic bugs that clear out with a game reboot, and maybe some of that even contributes to the success of some test launches, only to vanish in the next sitting.  Hard to say.

[Brainlord Mesomorph]

No I guess this is just somthing I do to myself. 

When I first started playing KSP I would build a rocket to completion in the VAB before I would put it on the pad, and then it would fall over.

Learning from that mistake, I started testing rockets earlier and earlier in the design process, these days, as soon as I put an engine, a fuel tank, a nose cone, and some fins together I stick it on the launch pad for quick test. 

Then I start adding parts…

And even though I’m testing periodically, apparently by the time I get the payload bay filled, and add the final things like solar panels, ladders, and RCS ports, I'm adding just enough weight and drag to ruin it, and *then*  I’m done for the night. 

Edited April 7, 2016 by Brainlord Mesomorph

[SpaceplaneAddict]

It's all in your launch profile. If you use THE EXACT same gravity turn, and thrust profile, there's no reason for it not to consistently make orbit. Test afew various gravity turns. Once you make orbit, note how much fuel you have left. That will be your margin for error within your launch profile. Just...

 

Stay Determined . . .

[numerobis]

Probably not what's affecting you, but sometimes I've had an issue where struts that were connecting fine in the VAB don't connect after reloading.

Otherwise, I recommend using MechJeb or GravityTurn or KOS to do the ascent. That makes it repeatable.

[ag3nt108]

Unless you update a KSP version a "working" design should work every time. As already mentioned though flying manually introduces a huge margin of human error.

[guitarxe]

Sometimes I noticed that if I click on "Revert To Launch", rather than "Rever to Hangar", then weird things will tend to happen afterwards, like some parts not working, or my whole staging UI disappearing and the like, which is invariably tied to the amount of mods I have, I'm sure, but it still might be different from a true "clean slate" if you go back to hangar and launch again from there.

[swjr-swis]

The tiniest changes to offset of a part that affects CoM or CoL can sometimes ruin what was a perfectly balanced craft just before, and it's hard to keep track of the changes one makes while building, even when one is in the habit of saving progressing versions of a craft.

I remember working on a tiny speedboat that at one point was balanced exactly right for a speed record run. I did several launches and speed runs and it was nice and repeatable, and I didn't even need to do anything other than apply throttle once it was in the water. I decided the boat itself was done, but the decoupling undercarriage that was supposed to bring it from the runway to the water needed a bit of work. Somewhere along working on that undercarriage, I changed something on the boat that I somehow managed not to save, and no amount of tweaking later got me the same results anymore.

The problem was obviously not pilot error or variability in control, since literally all I had to do was decouple/stage the undercarriage to sink away, and then apply throttle. Whatever it was I moved, it was enough to not allow it the same results as before.

It happens.

[Tw1]

Yep. Usually with launch vehicles.

I suspect human error though. But I wonder if the way pressure change due to temp change is stimulated could be a factor. 

[Jetski]

6 hours ago, numerobis said:

Probably not what's affecting you, but sometimes I've had an issue where struts that were connecting fine in the VAB don't connect after reloading.

Otherwise, I recommend using MechJeb or GravityTurn or KOS to do the ascent. That makes it repeatable.

This ^^  It has gotten me more than once, even with hidden struts

[Fleb]

Gremlins are real!

 

https://en.wikipedia.org/wiki/Kelly%E2%80%93Hopkinsville_encounter

[Slam_Jones]

Sounds like my spaceplanes...

"Hey, this one got the payload to orbit!  Awesome!  Guess that design is done for now."

*Comes back the next day*

*Can't leave atmosphere*

*Sad face*

[magnemoe]

This is mostly an atmosphere issue as I know, sometimes an rocket is just stable inside an narrow flight path who you was lucky to hit the first time. 
Spaceplanes are more vulnerable for this, however I had an minimal Tylo lander who worked twice: test and minimal grand tour, not later even in the same KSP version and all under mechjeb controll

[Mahnarch]

Sometimes there is a difference in your TWR and BAC from the night before.

 

I recommend increasing BAC and see what happens (take vids).

[n.b.z.]

I have long had a sneaking suspicion that the performance of a craft may somehow, somewhat, depend on whether it was just loaded from the assembly building, or loaded as a result of reverting to launch.

It really is just that - a suspicion - and I do not have any proof, numbers, etc.

Numerobis mentioned KOS. Well, this is why I got the above impression. A spaceplane ascent, 100% scripted from brake release to circularization, should be a completely repeatable exercise, and I seem to remember that this was somehow not the case, and I started to always load from the Hangar instead of reverting, "just to make sure".

I may very well be utterly wrong. In case I'm not, this may explain why developing and testing a craft (= reverting to launch a lot) may yield different performance than freshly loading your finished, brand new craft to take it on its first mission.

[Brainlord Mesomorph]

1 hour ago, n.b.z. said:

I have long had a sneaking suspicion that the performance of a craft may somehow, somewhat, depend on whether it was just loaded from the assembly building, or loaded as a result of reverting to launch.
...  I may very well be utterly wrong. In case I'm not, this may explain why developing and testing a craft (= reverting to launch a lot) may yield different performance than freshly loading your finished, brand new craft to take it on its first mission.

 

It ALMOST seems that way, right? But I can't imagine the game is THAT buggy. 

 

2 hours ago, Mahnarch said:

Sometimes there is a difference in your TWR and BAC from the night before.

I recommend increasing BAC and see what happens (take vids).

 

I'm watching the test flight through beer goggles?  "That plane seemed prettier last night" LOL

 

 

Oh And FYI everybody: 

My problem seemed to be drag. I had added lights, RCS blocks, antennae,  like 3 ladders, and it was the drag that broke the camel's back, costing about 175 m/s in dV.

Adding 10% more fuel to the 1st stage and she was good to go.

Edited April 14, 2016 by Brainlord Mesomorph

[Skorpychan]

Often. I blame the 'new' aero.

Of course, even before it, I used to get random failures for no adequately explained reason, or because I was a little sharp with the throttle.

[swjr-swis]

On 4/6/2016 at 0:56 AM, Brainlord Mesomorph said:

Does it seem to anyone else like a perfectly functional  ship you design one night simply doesn’t work the next day?

(...)

But it seems like five times now I designed something that makes orbit 3 or 4 or 5  times, and decide “yes that will do” and I’m done for the night, and then the next day the damn thing won't make orbit!

Now I have no idea what I’m doing differently,

 

A week and some beer goggles later:

44 minutes ago, Brainlord Mesomorph said:

My problem seemed to be drag. I had added lights, RCS blocks, antennae,  like 3 ladders, and it was the drag that broke the camel's back, costing about 175 m/s in dV.

 

Maybe that's where you should've started looking...

Problem solved, so all's well that ends up in orbit.

[Brainlord Mesomorph]

14 hours ago, swjr-swis said:

Maybe that's where you should've started looking...

Everything is always in the last place you look. 

[AeroGav]

What really drives me nuts is airplanes that constantly try to veer off to one side or other.    

My long range spaceplanes need 25 minutes to get to orbit, that's a long time to keep tapping Q since going one degree of course requires a costly orbital correction.     Worse, career mode games require you to fly to the other side of the Kerbin to take measurements - and i'm forced to keep tapping the keyboard the whole way.

I've added multiple vertical stabilizers,  used gull-wing dihedral, at best these can sometimes stop the bank angle exceeding 15 degrees so your plane only flies in a wide circle rather than diving to the ground inverted if left unattended for any period.   No, you can't use SAS because it's not designed for planes.  It keeps a constant nose angle that doesn't allow for magnetic heading changing as latitude increases,   and holds a constant pitch that doesn't allow for the curvature of the planet, so the nose slowly rises with time.

A minority of the time, these super stable designs actually do fly straight.  But the next time i use the same aircraft it veers left.   And the next time it wants to veer right.   If the plane is veering off course, quick save , quit game, and reload is the only way to cure it, which kind of confirms its a physics glitch.

 

[Vermil]

These tales seem suddenly familiar of late. I'll tell the story more completely elsewhere, when I'm ready, but there seem to be something very fishy going on with the physics simulation in v1.05.  Well, I'm frustrated. Very frustrated. Why are things not the same the next time? Why is a system prone to oscillations? And why, - why is energy fed into a such oscillating system? Amplifying it? When there is no such energy source present?

 

 

Edit:  Now I'm ready, and take most of it back. I've arrived at an explanation. Things were indeed NOT as before!  I had landed in a different region than assumed, with - apparently - different ground properties.

 

Still, there's a tendency to oscillations self-amplifying when they shouldn't. This also happens on the launch pad sometimes, with heavy rockets. This time I managed the landing by keeping the SAS and RCS active all the time (ironically, they could feed energy into the system), until "speed" finally expired into 0.0.  But I wouldn't be surprised if the oscillations start all over again if I quit the game and reload to continue at a later date. I hope I'm pessimistic about that, but just to be a bit safer, I'm going to continue playing without a break. 

Edited April 15, 2016 by Vermil

[Archgeek]

On 4/14/2016 at 0:57 PM, Mahnarch said:

Sometimes there is a difference in your TWR and BAC from the night before.

 

I recommend increasing BAC and see what happens (take vids).

It took me far, far too long to get that; largely due to my brain associating those letters with the BACC SRB before anything else. 

On 4/15/2016 at 0:40 PM, Vermil said:

Still, there's a tendency to oscillations self-amplifying when they shouldn't.

Yup, I nearly lost the craft in my avatar to a self-amplifying oscillation (actually, I lost it repeatedly but kept restoring it via powerful chronosorceries before stumbling desperately on the solution).  I'm tempted to put in a feature request for oscilation dampening, considering what really happens absent a viscous medium to carry energy away via sound is that large-scale, low-frequency flaperdoodles tend to decay into a high-frequency shipwide "ringing" which itself attenuates simply due to friction losses within the materials of vehicle.