Why does this vessel use RCS fuel this way?

[Zeiss Ikon]

I've encountered what seems an odd phenomenon, that has helped strand a Kerbal in Munar orbit, without a ship.

Here's a craft file: Selene 1 Mod 2

I launch this ship with RCS turned off, using main engine gimballing for control until staging.  Once I stage away the booster, fairing, and nose cone (above 70 km only due to a spectacular demonstration of the drag still present at 40+ km and 1000+ m/s), I turn on RCS so I can maneuver before igniting the second stage engine to circularize (the second stage also performs transMunar insertion and Munar orbit capture before being deorbited on the Mun).  This is where the odd thing happens.

It takes a lot of RCS authority to turn the remaining stack at this stage (initially, with full tanks in all modules), and the center of mass is far enough forward that the Lf/O thrusters on the second stage tank get a lot of help from the four sets of RCS quads on the lander and pusher stages.  However, the RCS on the lander and pusher stages don't consume fuel evenly from the command pods and RCS tanks, despite crossfeed being enabled on every decoupler and stack separator.  Rather, the lander's RCS fuel is depleted first, with the command pod being emptied before the stack RCS tank is used, and both of those being burned dry before any fuel is drawn from the pusher (top module at launch).

This doesn't seem to be a case of the game drawing propellant from the lowest tank first, which would make some sense for the usual staging order; rather, the command pod fuel is being depleted first.  The only solution I see is to remember to manually disable the command pod and stack tank feeds before launch, and then manually re-enable them before I'm ready to decouple the pusher and turn it around to dock with the lander (a la Apollo, which is the configuration I adopt before Munar orbit insertion so the pusher is ready to maneuver itself and the lander as soon as the second stage is jettisoned).

Am I missing something relative to propellant consumption order?  I think this is the first time it's really matter which tank empties first -- but this time, it led to a Kerbal launching from Mun without enough RCS propellant left to assist when the main engine's tanks ran dry before establishing orbit.  The pusher pilot, meanwhile, was sitting in a safe orbit, with a nearly full RCS tank, and no way to help...

Edited January 29, 2017 by Zeiss Ikon
Add screenshot of vessel

[Snark]

Moving to Gameplay Questions.

A suggestion:  Posting a link to your .craft is fine, but could you post a screenshot, as well?  For most questions, you're actually more likely to get a helpful response with a screenshot than with a .craft file.  Reason:  Lots of problems are easily diagnosed from a screenshot by an experienced eye, and can be done very easily by just reading the forum post.  If all you've provided is a .craft file, then you're requiring any potential helpful question-answerer to download your file, save it into their KSP game, spin up KSP (which requires that they have access to their KSP computer), and fly the ship to an appropriate situation to reproduce your problem.  All of that is pretty inconvenient and will discourage a lot of folks from bothering to try to answer the question.

For example, I'm not at my KSP computer right now, so your .craft file is completely useless to me.  If there were a screenshot, I could probably offer much better advice than the guesswork you'll see below. 

Anyway, on to your questions.

22 minutes ago, Zeiss Ikon said:

despite crossfeed being enabled on every decoupler and stack separator

Note, this is irrelevant.  RCS is "whole-ship" flow, unlike LFO engines.  All RCS thrusters have access to all monopropellant on the ship, regardless of crossfeed status.

22 minutes ago, Zeiss Ikon said:

Am I missing something relative to propellant consumption order?

Have you checked the flow priority on your monopropellant tanks, and compared that with the flow priority on the lander?  If you want the lander's monopropellant to be used last, you want to make sure it has the lowest flow priority.

Usually the game does a pretty good job of setting up the flow priorities by default so that it "just works" and you don't have to futz with it, and it drains the way you'd like (i.e. lowest stages first), but depending on how your ship is constructed and the order in which you put stuff together in the VAB, the default setup may not be working right for you.  You can manually adjust the flow priorities, both in the VAB and in flight.

(I think you may need to have "advanced tweakables" turned on to be able to do that, though.  Sorry, I always have advanced tweakables turned on myself, so I'm not well-versed in which features require that and which don't.)

[cantab]

The game's fuel flow rules aren't really designed with Apollo-style missions in mind. As KSP sees it your orbiter is the main ship and the lander the previous stage. Though I'm not sure why the lander pod goes before its main tank.

You can fix it by enabling Advanced Tweakables then setting the flow priority on the RCS tanks to be all the same or otherwise as desired.

[pincushionman]

Does momoprop consumption even obey flow priority? Or does it just draw from all tanks, period?

If it's the latter, and your lander tank is smaller than your xfer stage's, it will deplete first because it pulls MP from both tanks at the same rate.

[FullMetalMachinist]

19 minutes ago, pincushionman said:

Does momoprop consumption even obey flow priority?

Yes, and so does electric charge consumption with batteries. 

[Snark]

1 hour ago, pincushionman said:

Does momoprop consumption even obey flow priority?

Yup.  All resources do, including electricity (as @FullMetalMachinist points out).  Even works for modded resources, such as life support supplies.

It's actually cool that it does that-- gives the player a lot of flexibility in ship design.  For example, suppose I expect to need a lot of electricity on reentry, because my reentry vehicle is somewhat aerodynamically unstable and I need to use the reaction wheels a lot to hold retrograde.  I can set things up so that its batteries (including the built-in ones in the command pod, if there is one) have a lower flow priority than those on the rest of the ship that gets staged away before or during reentry.  That way, I'm assured that I have full batteries for the process.

Another example of usefulness:  Suppose I'm playing with TAC Life Support.  There are some parts available for that mod for reprocessing waste into usable resources (e.g. processing CO₂ back into oxygen, or waste water into water).  These parts typically have some storage space for whatever waste they're designed to process.  I like to set them with really high flow priority.  Thus, for example, suppose I have a space station with a recycler on it, and I have some ships "parked", docked to the station:  this guarantees that all the waste from the ships' crews will go straight into the recycler, rather than filling up the limited waste storage capacity on the ships.

[Zeiss Ikon]

Flow priority -- AHA!!  Must go turn on advanced tweakables...

Actually, I'd like the flow to be equal from both modules' stack tanks, but the command pod tanks last so there's RCS fuel left for reentry on the lander's pod.  Sounds like I should be able to accomplish that by setting the command pod tanks to lowest priority.  BTW, the next upgrade on this design will likely be swapping the lander and return pusher positions, so the lander can undock, land and return and leave the return pusher still coupled to the second stage.  By doing that, I should be able to delete the pusher's own Lf/O tank and engine entirely, upgrade the lander's Lf/O tankage and engine to give enough dV and thrust to land on Duna, and still save fuel (via saving mass) enough to make the Duna trip, or possibly even Dres.  Ike and Gilly are a given if I can make Duna, of course...

10 hours ago, Snark said:

Moving to Gameplay Questions.

A suggestion:  Posting a link to your .craft is fine, but could you post a screenshot, as well?

Sorry, I intended to, but I had two power drops in ten minutes (heater flipping a breaker) and by the time I had that straightened out, I'd forgotten to take and post a screen shot.  A pad shot won't really help here anyway, all the relevant parts are inside the fairing, but I might already have a shot of the second stage with the lander and return stage still docked.  If so, I'll edit it in above.

Edited January 29, 2017 by Zeiss Ikon
Change "docked" for "coupled"