Goddard-Problem: Maximal launch altitude Challenge

[mhoram]

Thanks for all the new entries!

I have the greatest respect for all of you who push the boundaries even further and appreciate the effort you put into this.

@Sando Mutt: I tried your script and came in several tries to altitudes of ~61,6km which also differs from Padishars results of your script.

@Padishar: Your script brings me to an altitude of higher than 61,9km and at one try I was at 61,992km.

These differences in the results are a great problem for me, because I want to keep the challenge fair and comparable but currently the results are not comparable.

That is why I went to seek help on the kOS-thread. I hope, that they can help to figure out a way to keep the results consistent.

I will update the leaderboard as soon as I have a bit more clarity.

On a different matter I would like to give an overview of the entries so far:

We had some entries that have no or just a little description of the ascent path.

Some entries throttle down just below mach 1 for some time and accelerate again afterwards.

A few do the same just below mach 2.

And Nao and Padishar provided entries that throttled down twice just below mach 1 and mach 2.

My gut feeling says that the last variant should probably be the most efficient of them. I wonder if a third throttling-down below mach 3 would make sense.

[Sando Mutt]

My current understanding of the issues with repeatability of launches is that they are related to aerodynamics more than kOS engine. In my tests I have used my script at different hours of the day, and the results are greatly different, more than 500m. of difference; if you repeat the launch at the same time, each one will be different also, but the interval is well inside 100m. between the lowest to the highest altitude. I do not know the details of the stock aerodynamical model, but my best results occur at 5:00h Kerbin time, which is also the hottest hour. So I tempted to consider that the aero model takes into account temperature of air, increasing and decreasing respectively the density of the atmosphere, which of course affects drag and thus flight profile.

Please, try then my script at just that hour, to see if that is the "problem".

Also I have tested Padishar script 6 times at 5:00h Kerbin time, my results:

1. 61986
   
2. 62023
   
3. 62023
   
4. 62011
   
5. 62002
   
6. 62017
   

And at 2:05h Kerbin time, the coolest hour:

1. 61369
   
2. 61378
   
3. 61404
   
4. 61437
   
5. 61432
   
6. 61422
   

I find that values totally in line with my results. So I think this validates my new entries, Padishar ones and mhoran tests, and makes Padishar script a really good one, capable of reaching 62000m. solidly. Congrats man, try it at 5:00H and post your result to update the leaderboard.

To finish, a new entry in manual mode 62096m.:

Just the same flight profile as in my previous entry, but with a very adjusted timing in the thrust changes. Launch time: 5:00h Kerbin time.

[mhoram]

@Sando Mutt: Thanks for your indepth analysis.

KSP 1.0 introduced a Temperature-model:

Both lift and drag are dependant on density and the speed of sound; both properly calculated from temperature and pressure.

Atmospheric temperature, and thus density, takes latitude and sun position into account.

I was aware of that but did not expect the temperature to have such a large impact on the results of this challenge.

I find it a bit counterintuitive, that 5:00h Kerbin is the hottest time at Kerbal Space Center, because it is before midday - while in reality the hottest time is in the afternoon.

For reference I made most of my flighttests at around 5:45 Kerbin Time.

I will test the scripts at 5:00h Kerbin time and try other times and make a note about these findings in the OP.

Also I got an answer from the kOS side.

My first guess would be FixedUpdate rates. Despite the name "Fixed" they're not really Fixed. Given enough of a load on your gaming computer, one run can have a different physics delta time than another.

A far better name for FixedUpdate, that was more honest about what it really is, would be "CappedUpdate". It's an update that you can cap at a maximum rate, but not one that's guaranteed to actually always be fixed at that rate. If your system is under heavy enough load, Unity WILL slow down your "Fixed" updates so your physics delta time is longer than you configured it for. What that configuration option *Really* does is say "even if my computer is running fast and not being taxed too hard, don't allow physics updates to be any FASTER than this, but you can still make them SLOWER than this if you have do.".

Without knowing the actual scripts in question, it's hard to give a more definitive answer to what's going on.

This sounds like the reason for the <100m differences when launching at the same time.

[Padishar]

I'll try running mine at 5:00 later today.

The performance of the computer also definitely plays a part. On my main machine that maintains the limit of 60fps constantly during the test I get significantly better results than I do on my old laptop that only manages about 19fps (and that needs me looking at the sky). I'm pretty sure that they are launching at the same time of day because I created a new sandbox on both and the times are still on 0 (well, probably a minute or two). I'm going to knock up a little plugin that draws a graph of the physics delta time value for each frame (and probably the real time between FixedUpdate calls).

I do know that it only takes a very small difference in vertical speed at 20km to make a significant difference in final altitude. There is a bit in my script that dumps this out as it is handy for getting a quick idea of how high you will get...

[mhoram]

@ Padishar:

I find it sad, but believe that we have to life with the fact that the computer-performance has an impact on the results. In this regard the approach that was made in the game Supreme Commander would have been better, where the computing power has no impact on what happens in a single game-time-frame - it slowed the game down but was consistent over all platforms.

For the plugin idea, have a look at:

http://forum.kerbalspaceprogram.com/threads/116034#PhysicalTimeRatioViewer

- - - Updated - - -

So ... some new findings.

I used the AeroGUI mod to figure out, when the temperature is the hottest at the launchpad. It reaches its maximum roughly at the time when the sun stands at 45° above horizon in the morning - this also means that the best launchtime is not the same all days, but shifts from day to day.

On the first day it is near 5:00h

Even at that best launchtime I come only to ~61,6km with Sando Mutt's 62.116km-script.

The results with Padishars-script are with 62,010 km(+-) in the same range that Sando Mutt reached.

And the script I made for adamgerd also reaches 61,6km.

Slowly I run out of ideas for the cause of the discrepancies.

Edited August 26, 2015 by mhoram

[Padishar]

I gave mine various tests at different times and I agree that the best time is 05:00 but I couldn't get it much over 61.7km. After making a few little tweaks I got it to 61867:

Tweaked script. Given this one goes 150m higher for me than my last one, I would be interested to see how high it goes for you folks...

As for possible causes for the discrepancies, what mods do you have installed? I only have kOS, KER and MM in this install at the moment and have checked ModuleManager.physics is identical to Physics.cfg so nothing is messing with it. Also, what hardware/os/video settings/physics delta are you using?

[mhoram]

Tweaked script. Given this one goes 150m higher for me than my last one, I would be interested to see how high it goes for you folks...

My three tests with your script resulted in: 62,085; 62,111; 62,126

As for possible causes for the discrepancies, what mods do you have installed? I only have kOS, KER and MM in this install at the moment and have checked ModuleManager.physics is identical to Physics.cfg so nothing is messing with it. Also, what hardware/os/video settings/physics delta are you using?

Mods: AeroGUI, KER, kOS, MJ, MM, Telemachus, Toolbar, my goddard.cfg from the OP.

PC: Win 8; i7: 2,60GHz

Settings: Default (physics-delta: 0.04) + changed graphics to fullscreen.

Yes, MM.physics has the same values, I also verified a few values via alt-f12.

---

From now on I will also allow f3-screenshots after reaching apoapsis for determining the final altitude (will update the leaderboard for the entries that are affected by this change).

Edited August 27, 2015 by mhoram

[Padishar]

You appear to be using a strange sort on the leaderboard...

[mhoram]

"Strange Sort" is a new kind of sorting algorithm that includes the functionality of delayed sorting and external input. I will publish it soon as a new Wikipedia-article

Edited August 27, 2015 by mhoram

[Aaronfinch123]

It has been a long time (V0.16) since someone brought up the Goddard problem here on the challenges-forum. So I felt it is time with the new aerodynamic system to have a new take on it.

Mission Description

Launch the provided rocket "Triple-G" from the launchpad as high as possible into the upper atmosphere.

http://i.imgur.com/XGGd5Df.jpg

Scoring

The score you get is the altitude in meter as shown in a screenshot you provide. The higher the better.

Restrictions

• Use this stock rocket (Triple-G.craft) without changing it (thrust limiter adjustments are allowed)
  
• No cheating, config file changes or debug console
  
• No gui-altimeter or physics changing addons
  
• KSP V 1.0.4
  
• Please verify that you have an original stock physics.cfg (available in this thread)
  

In order to be able to use MechJeb, Telemachus or kOS, I provide an officially sanctioned module manager config file (goddard.cfg)

{
MODULE
{
name = kOSProcessor
diskSpace = 5000
}
MODULE
{
name = TelemachusDataLink
}
MODULE
{
name = TelemachusPowerDrain
powerConsumptionBase = 0.01
powerConsumptionIncrease = 0.02
}
MODULE
{
name = MechJebCore
MechJebLocalSettings {
MechJebModuleCustomWindowEditor { unlockTechs = flightControl }
MechJebModuleSmartASS { unlockTechs = flightControl }
MechJebModuleManeuverPlanner { unlockTechs = advFlightControl }
MechJebModuleNodeEditor { unlockTechs = advFlightControl }
MechJebModuleTranslatron { unlockTechs = advFlightControl }
MechJebModuleWarpHelper { unlockTechs = advFlightControl }
MechJebModuleAttitudeAdjustment { unlockTechs = advFlightControl }
MechJebModuleThrustWindow { unlockTechs = advFlightControl }
MechJebModuleRCSBalancerWindow { unlockTechs = advFlightControl }
MechJebModuleRoverWindow { unlockTechs = fieldScience }
MechJebModuleAscentGuidance { unlockTechs = unmannedTech }
MechJebModuleLandingGuidance { unlockTechs = unmannedTech }
MechJebModuleSpaceplaneGuidance { unlockTechs = unmannedTech }
MechJebModuleDockingGuidance { unlockTechs = advUnmanned }
MechJebModuleRendezvousAutopilotWindow { unlockTechs = advUnmanned }
MechJebModuleRendezvousGuidance { unlockTechs = advUnmanned }
}
}
}

@PART[probeStackSmall]

Submissions

A valid submission must contain at least the following information

• A screenshot depicting the rocket at it's highest possible altitude or a F3-readout after reaching apoapsis.
  
• A list of addons you used, that helped with the launch (readouts, steering, thrust)
  
• Optionally it would be great if you would provide stats about the ascent (Telemachus for example) or infos on how you reached the altitude
  

Leaderboard

1. 61867: Padishar (kOS-script)
   
2. 61761: Nich (manual launch)
   
3. 61704: Sando Mutt (manual launch)
   
4. 61525: Gaarst (manual launch)
   
5. 61503: Nao (manual launch)
   
6. 61478: MadChris48 (manual launch)
   
7. 61296: Padishar (manual launch)
   
8. 61194: SpaceWeaver (manual launch)
   
9. 61175: Feradose
   
10. 61137: mhoram (kOS-script)
    
11. 58824: kookoo_gr (manual launch)
    
12. 58330: Foxster (manual launch)
    
13. 56969: Cool791
    
14. 53914: Kaboom!
    
15. 49261: mhoram (manual launch)
    
16. 46070: littlebuddy0
    

The Attic - the place where you beat yourself

1. 61604: Padishar (kOS-script)
   
2. 60915: Padishar (manual launch)
   
3. 60883: MadChris48 (manual launch)
   
4. 59963: MadChris48 (manual launch)
   
5. 59531: MadChris48 (manual launch)
   

Pending Verification

1. (to be verified: 62891: adamgerd) (manual/MJ launch)
   
2. http://forum.kerbalspaceprogram.com/threads/128509-Goddard-Problem-Maximal-launch-altitude-Challenge?p=2158468&viewfull=1#post2158468
   
3. http://forum.kerbalspaceprogram.com/threads/128509-Goddard-Problem-Maximal-launch-altitude-Challenge?p=2159963&viewfull=1#post2159963
   

Since it was requested, participants may use this badge:

http://i.imgur.com/C0mg0CA.png (large version)

Does this include adjusting the thrust limiter in VAB?

Edited September 8, 2015 by Red Iron Crown

[Padishar]

There wasn't really any need to quote the whole of the OP, especially considering it clearly states that thrust limiter adjustments are allowed in the first line of the restrictions...