Sideways slope of KSC runway

[Padishar]

Apologies if this is widely known but I don't recall ever seeing it mentioned anywhere and it hasn't occurred to me before now. While doing testing of the new slope readout I have just added to KER, I discovered something about the KSC that sheds some light on why craft misbehave on the runway.

 

So, the runway has a slight slope to the right along the whole of the length. The heading of the "downhill" direction varies from 111° at the west end of the runway to 247° at the other end. Thinking about it, this must be because all of the "flat" concrete at KSC is parallel and, due to the curvature of the planet, it is only truly horizontal at one point somewhere among the buildings. The has the result that gravity doesn't quite act vertically downwards anywhere on the runway, it always pulls slightly to the right and either forward or backward depending on which end you are at (basically always slightly towards the buildings). This explains why planes and rovers parked at the end of the runway will slowly start rolling along the runway when the brakes are turned off. The same thing happens on the launch pad for exactly the same reason. KER shows 0° slope on the launch pad but it only shows 1 decimal place and there is definitely still a slope. If you go to the far point of the concrete ring road around the launch pad then the slope is just enough to show as 0.1°.

Edited April 20, 2016 by Padishar
Fixed album and degree symbols

[Russoft]

The runway is perfectly flat. Kerbin is curved. Hence at the edges of the runway you're further from the surface of Kerbin.

It helps if I read the post before replying. I see that you understand what's going on

Edited June 25, 2014 by Russoft

[Drew Kerman]

I knew and understood the runway slope but did not realize that was helping wheeled craft turn to the right as well. I always thought the right turning was from slightly-misaligned gear.

[I_Killed_Jeb]

please have some rep

[problemecium]

I've suspected something like this ever since hearing Scott Manley's anecdote about the Space Shuttle runway and how KSP's runway is perfectly flat and thus pitched up at either end relative to the curvature of Kerbin, whereas the real Space Shuttle runway had to be made to closely follow Earth's curvature.

Also KSC is not exactly on the equator, so there's probably a bit of Coriolis force at work here.

[RoboRay]

Yeah, the long, shallow "dip" down the length of the runway due to it being perfectly flat and not conforming to the curved surface beneath it has been talked about since shortly after the runway was first introduced...

But the "banking" due to the plane of the runway being tangent to the spherical surface well off to the side of the runway is a new concept, I think. Very interesting revelation!

Edited June 26, 2014 by RoboRay

[Padishar]

Yeah, the long, shallow "dip" down the length of the runway due to it being perfectly flat and not conforming to the curved surface beneath it has been talked about since shortly after the runway was first introduced...

But the "banking" due to the plane of the runway being tangent to the spherical surface well off to the side of the runway is a new concept, I think. Very interesting revelation!

I presume the runway was introduced before the great forum wipe which would explain why I couldn't find much about it. I've only been playing since just before 0.23 was released and I've seen plenty of people talking about planes drifting right on the runway but they've always been told their landing gear isn't straight, probably because of bugs in the SPH symmetry handling. I'm not actually claiming that this slight slope to the south is definitely responsible for any southward drift but it makes more sense to me than the wheels being out of alignment. The code to generate symmetrical transforms for the parts when attached is not complicated and any errors should be down at the limits of the floating point representation and shouldn't be noticeable.

Edited June 26, 2014 by Padishar
removed a spurious "a"

[NecroBones]

Hah, yeah I knew about the sloping toward the center of the runway, due to it being flat... but I didn't realize the entire KSC complex was like that, and thus the runway slopes to the right. Good find!

[Laie]

This explanation makes a lot of sense, but...

Why do I have to point slightly to the right on the runway? I found that I need to maintain a course of 90.3 degrees. 90.2 veers off to the left, 90.4 to the right.

With the runway banking right, one should expect that pointing left was the way to go.

Am I missing something?

[RoboRay]

I've seen plenty of people talking about planes drifting right on the runway but they've always been told their landing gear isn't straight, probably because of bugs in the SPH symmetry handling. I'm not actually claiming that this slight slope to the south is definitely responsible for any southward drift but it makes more sense to me than the wheels being out of alignment. The code to generate symmetrical transforms for the parts when attached is not complicated and any errors should be down at the limits of the floating point representation and shouldn't be noticeable.

Part of the problem of wheels not being straight isn't due to them not being straight in the SPH when built... it's when a load is applied when the physics engine kicks in and joints between parts start to flex. If the flexing isn't perfectly symmetrical, you can end up with one wheel angled slightly differently than another wheel.

And if the runway is sloped to one side, asymmetrical flexing becomes much more likely. I think your discovery is a real eye-opener into what's happening.

Edited June 26, 2014 by RoboRay

[Padishar]

That's a good question, but I can't give you a definite answer. First off, that sequence of pictures is not demonstrating drift on the runway, it is simply demonstrating the slope via the KER window. Second, like I said in #7, I'm not even claiming that this is definitely responsible for drift, let alone is the only thing that affects it. If it were only the slope that affected it then it wouldn't make any difference what course you set (within reason), it would always drift down the slope to the right. You would need to go further south than 111° at the start of the runway to make the slope go to the left and then you will leave the runway pretty quickly (you would be pointing at the buildings).

The angle and direction of the slope is not constant along the runway so there could be other effects involved. At the start your plane is pitched down and banked to the right very slightly. As you go along the runway you pitch up gradually due to the slope changing but the roll to the right stays much more constant. The "aerodynamic" forces (if you can call the stock behaviour that while keeping a straight face, or you use FAR) will be changing as you move along the runway.

There are also other definite inaccuracies in the physical modelling because simply releasing the wheel brakes on that stationary rover can noticeably change the direction it is pointing in.

Also, despite what I said about the symmetrical transforms being simple, this doesn't mean they are implemented well. It is quite possible that there is some design quirk of your vessel that is overwhelming the sideways force caused by the banked runway.

I guess my main reason for posting this thread was to prompt people to discuss it and maybe perform more detailed experiments with very carefully assembled test craft.

One possible thing for Squad to do would be to modify the KSC model so that the origin point of it is on the center line of the runway rather than somewhere among the buildings (and then move the whole model on the planet to compensate). This would remove the sideways part of the slope on the runway but wouldn't affect the "dip" along the length. To fix the dip as well would require significantly more work, either designing the model with the runway in sections pre-wrapped to the curvature (though this would cause issues for RSS etc) or by programmatically modifying the polygons for the runway surface after the model is placed on the planet, e.g. the runway model would be made with lots of small strips but still be perfectly flat and when it is placed on the planet the strips would be moved to follow the curvature. The trouble is, without further detailed experiments, it is not known whether this would be any improvement or if the "joins" in the runway would cause other issues (e.g. a big plane driving over the little bumps might rip the wheels off).

Edit: RoboRay, yes, joint flexing under load (which is changing all the time due to thrust, lift and drag changing) must play quite a large part too and the slope would imply more asymmetrical flexing...

Edited June 26, 2014 by Padishar