Difference between revisions of "Talk:Tutorial:Aeris 4A mission"

From Kerbal Space Program Wiki
Jump to: navigation, search
(Some efficiency numbers for the ascent profile, and a challenge to improve on them.)
(1.0 ascent profile: new section)
Line 12: Line 12:
 
:: The challenge is open to anyone that feels motivated to try some different ascent profiles (I recommend primarily using the Aerospike for the sake of sanity on takeoff, then verifying promising profiles with the LV-T30).  I may well play around with it a bit more, but that's probably enough for now, I think.  Anyone that's unhappy using MechJeb can test by just looking at the fuel remaining once you achieve 100km circular orbit.
 
:: The challenge is open to anyone that feels motivated to try some different ascent profiles (I recommend primarily using the Aerospike for the sake of sanity on takeoff, then verifying promising profiles with the LV-T30).  I may well play around with it a bit more, but that's probably enough for now, I think.  Anyone that's unhappy using MechJeb can test by just looking at the fuel remaining once you achieve 100km circular orbit.
 
:: --[[User:Murph|Murph]] ([[User talk:Murph|talk]]) 20:51, 15 November 2013 (CST)
 
:: --[[User:Murph|Murph]] ([[User talk:Murph|talk]]) 20:51, 15 November 2013 (CST)
 +
 +
== 1.0 ascent profile ==
 +
 +
Ok, so after a little testing, I can confirm that the pre-1.0 tutorial is most certainly obsolete as a whole.  A continuous 30° ascent profile just doesn't work for the 1.0 atmospheric physics.  Heat isn't actually the major problem; the bigger issue is getting enough speed before switching to rocket mode, so as to have sufficient fuel to achieve stable orbit.  I think it's probably possible, just needs far more effort, and a much more complicated ascent profile.  I have successfully achieved a 75x75km orbit in 1.0.2 with a modified Aeris 4A, but not yet with the stock one.  I have been trying both stock and modified to try to figure out just what's needed.  Overall, it's now much more difficulut.
 +
 +
My current thoughts on ascent profile:
 +
 +
* Pull up extremely gently on the runway, as the plane has a habit of pitching up quite violently as soon as it's "unstuck".  Gentle control inputs do work to give a nice takeoff.
 +
* Just go immediately to 60° above horizon after takeoff, as the new turbojet performance gives quite extreme power at lower altitudes, once you've a reasonable air speed (150 m/s and above).  Use that power to get out of the lower atmosphere ASAP.
 +
* At about 7,500m, gently drop the nose from 60° to 45° above horizon.  The thickest air is below us now, let's start gaining speed as well as altitude.
 +
* At about 12,500m, gently drop the nose from 45° to 30° above horizon.  We need to start slowing the gain of altitude, to focus on speed.
 +
* At about 15,000m, gently drop the nose from 30° to about 15°, or maybe even 10°.  Altitude is no longer a priority, it's speed, and we need to build up to over mach 3.0 (1020 m/s) before switching to rockets
 +
* At about 20,000m, the initial ascent is basically done, and it's all about speed.  Keep the nose between 0° and 10°, gaining speed (ideally without losing altitude).  Be careful to avoid big control inputs, as they scrub speed off significantly.
 +
* Somewhere between 20,000m and 25,000m, with speed at least mach 3.0 (1020 m/s), we need to switch to rocket mode.  This is the tricky spot right now, trying to find the perfect combination of speed (and actually not just speed, but the prograde vector; angle matters now), altitude, and nose angle above horizon.
 +
 +
That's it for now.  I'm continuing to work on it.  Educated / intelligent input on the fine details of the ascent profile is welcome, although I don't really need random "have you tried xxxx?", unless you have some golden not-a-cheat trick which vastly improves things.  Ideally we're trying to do this without any modifications to the craft itself.
 +
 +
--[[User:Murph|Murph]] ([[User talk:Murph|talk]]) 17:56, 3 May 2015 (UTC)

Revision as of 17:56, 3 May 2015

I've tried following this tutorial, and found I had better luck with a (much) higher initial rate of climb, and then once near the flameout threshold, a liquid rocket engine climb 10-15 degrees above the prograde. Did they change some of the dynamics since this tutorial was created? Thoughts? — Preceding unsigned comment added by The braughtwurst (talkcontribs)

Yes, I found that dropping the nose to just 10° @ 15,000m resulted in it being stopped from reaching 20,000m by chronic flaming drag somewhere around 17–18,000m. I don't know if this is something which has changed across KSP versions, as I've only attempted this tutorial under 0.22, but I'd guess that it probably did work ok when written, with an older KSP version.
Personally, I found that going for 30° all the way from just after takeoff until a 100km apoapsis is achieved works quite nicely, and seems fairly efficient. I have given the tutorial a fairly major overhaul, added lots of details, etc, and believe it now represents something fairly easily achievable in 0.22.
It's quite possible that there's a better combination of angles for maximum efficiency, but I put my efforts into testing if 30° was simple, easy, reasonable, repeatable, achievable, etc; rather than seeking absolute maximum efficiency. Without doing another test right now to confirm, I seem to remember that MechJeb tells me that I've got around 400m/s dV remaining after circularising at 100km (plenty for a little manoeuvring, then de-orbit). I'll go ahead and say that 30° is NotBad™, but am more than happy for someone to improve on my improvements. ;-)
--Murph (talk) 17:13, 15 November 2013 (CST)
Ok, I had a spare moment or 2 and did a couple of quick tests to get some basic numbers on efficiency for a straight 30° pitch above horizon from runway up to 100km circular orbit, with jet to rocket transition at 20km. Using MechJeb for numbers, to hold a perfect 30° angle all the way up, and make doing a perfect circularisation burn easy, but nothing fancier than that.
The stock Aeris 4A has a little over 300m/s Δv remaining. A pleasant surprise was the stock craft with just the LV-T30 replaced by the Aerospike, which had a little over 500m/s Δv remaining. In both cases, 100km apoapsis was achieved above 50km altitude, meaning that there is relatively negligible atmospheric drag loss during the coast up to apoapsis. Also, in both cases, speed from runway up to 20km steadily tracked at approx 110–120% of terminal velocity, so 30° seems very close to optimal to me for that entire phase of the flight. Shallower would be slowing the ascent too much from drag, steeper would be slowing it too much from fighting gravity and probably losing some of the benefit of lift from the wings, I think.
So, I believe that the question is mostly about the optimal ascent profile above 20km altitude, where terminal velocity essentially becomes a non-issue. For the purpose of this article, it needs to be something fairly easy to hand-fly, without too many precise changes, and a forgiving error margin. It also needs to offer significant improvement over just holding 30°, really, to be worthwhile. The primary burn should probably end at or above 50km altitude. Ideally, it should work well for both the LV-T30 and the Aerospike.
The challenge is open to anyone that feels motivated to try some different ascent profiles (I recommend primarily using the Aerospike for the sake of sanity on takeoff, then verifying promising profiles with the LV-T30). I may well play around with it a bit more, but that's probably enough for now, I think. Anyone that's unhappy using MechJeb can test by just looking at the fuel remaining once you achieve 100km circular orbit.
--Murph (talk) 20:51, 15 November 2013 (CST)

1.0 ascent profile

Ok, so after a little testing, I can confirm that the pre-1.0 tutorial is most certainly obsolete as a whole. A continuous 30° ascent profile just doesn't work for the 1.0 atmospheric physics. Heat isn't actually the major problem; the bigger issue is getting enough speed before switching to rocket mode, so as to have sufficient fuel to achieve stable orbit. I think it's probably possible, just needs far more effort, and a much more complicated ascent profile. I have successfully achieved a 75x75km orbit in 1.0.2 with a modified Aeris 4A, but not yet with the stock one. I have been trying both stock and modified to try to figure out just what's needed. Overall, it's now much more difficulut.

My current thoughts on ascent profile:

  • Pull up extremely gently on the runway, as the plane has a habit of pitching up quite violently as soon as it's "unstuck". Gentle control inputs do work to give a nice takeoff.
  • Just go immediately to 60° above horizon after takeoff, as the new turbojet performance gives quite extreme power at lower altitudes, once you've a reasonable air speed (150 m/s and above). Use that power to get out of the lower atmosphere ASAP.
  • At about 7,500m, gently drop the nose from 60° to 45° above horizon. The thickest air is below us now, let's start gaining speed as well as altitude.
  • At about 12,500m, gently drop the nose from 45° to 30° above horizon. We need to start slowing the gain of altitude, to focus on speed.
  • At about 15,000m, gently drop the nose from 30° to about 15°, or maybe even 10°. Altitude is no longer a priority, it's speed, and we need to build up to over mach 3.0 (1020 m/s) before switching to rockets
  • At about 20,000m, the initial ascent is basically done, and it's all about speed. Keep the nose between 0° and 10°, gaining speed (ideally without losing altitude). Be careful to avoid big control inputs, as they scrub speed off significantly.
  • Somewhere between 20,000m and 25,000m, with speed at least mach 3.0 (1020 m/s), we need to switch to rocket mode. This is the tricky spot right now, trying to find the perfect combination of speed (and actually not just speed, but the prograde vector; angle matters now), altitude, and nose angle above horizon.

That's it for now. I'm continuing to work on it. Educated / intelligent input on the fine details of the ascent profile is welcome, although I don't really need random "have you tried xxxx?", unless you have some golden not-a-cheat trick which vastly improves things. Ideally we're trying to do this without any modifications to the craft itself.

--Murph (talk) 17:56, 3 May 2015 (UTC)