Fuel Flow Rules (0.24.2)

[Padishar]

Thanks very much for the crafts. I'll grab them in a bit when I release an update for KER that will include this fix along with numerous other fixes and optimisations. Hopefully I will be able to get an official answer about it...

[Kasuha]

Experiment: docking may change root part and order in which monopropellant tanks are used

Command pod is root part.

With initial setup, bottom tank is drawn first because it is behind a decoupler.

When I reattached the "hose" to the bottom docking port, ordering has changed and upper tank is drawn first because now that is behind a decoupler relative to the root part which stays to be the command pod.

After I interrupted the "hose" in the middle, what remains of former connection is the new root part and upper tanks continues to be drawn first.

After I undocked the connection, root part was restored (to the HubMax part) and now bottom tank is drawn first again.

Craft file

Javascript is disabled. View full album

24-Aug-2014: verified for 0.24.2

Edited August 24, 2014 by Kasuha

[Kasuha]

Experiment: disabled monopropellant tanks are skipped; when enabled, depleting returns to their level

Command pod is the root part. One of monopropellant tanks is disabled. Monopropellant is drawn from other tanks according to their priority. As soon as the tank is enabled, drawing returns to its level and tanks on lower priority stop being used until it is depleted.

Craft file

Javascript is disabled. View full album

24-Aug-2014: verified for 0.24.2

Edited August 24, 2014 by Kasuha

[Kasuha]

I have cleaned up all my testing crafts and updated most of experiments with testing craft files. They're not always the same craft but they're equivalent, usually simpler to use and more telling. There are two experiments waiting to replicate for liquid fuel/oxidizer drawing rules.

[Kasuha]

Experiment: intake air distribution

Probe core at the top is the root part, hub modules were placed in tree-like manner, i.e. first straight from the core, then branching to sides. This order apparently plays no role.

25 engines were mounted to hub modules in lines, starting in upper left corner horizontally and ending in bottom right corner.

There are several test crafts involved in this experiment, differences being in how three air intakes were placed in between engines.

In the first case, air intakes were placed first, then all engines. Notice that engines are satisfied in order of placement.

Craft file

In the second case, air intakes were placed after first row of engines. Notice that engines are running starting with second row.

Craft file

In the third case, air intakes were placed before last row of engines. Notice that not only last row, but also first row now has engines running. This means the air distribution loops over to the beginning of the ship after reaching the end.

Craft file

In the last case, one intake was placed before second row, and two intakes were placed before fourth row. Notice that in second row, less engines are running than in fourth row, indicating the second row only uses intake air from the first intake.

Craft file

24-Aug-2014: verified for 0.24.2

Edited August 24, 2014 by Kasuha

[Kasuha]

Experiment: electric charge distribution

Probe core at the top is the root part, hub modules were placed in tree-like manner, i.e. first straight from the core, then branching to sides. This order apparently plays no role.

25 ion engines were mounted to hub modules in lines, starting in upper left corner horizontally and ending in bottom right corner.

There are several test crafts involved in this experiment, differences being in how three solar panels were placed in between engines.

In the first case, solar panels were placed first, then all engines. Notice that engines are satisfied in order of placement.

Craft file

In the second case, solar panels were placed after first row of engines. Notice that engines are running starting with second row.

Craft file

In the third case, solar panels were placed before last row of engines. Notice that not only last row, but also first row now has engines running. This means the electricity distribution loops over to the beginning of the ship after reaching the end.

Craft file

In the last case, one solar panel was placed before second row, and two panels were placed before fourth row. Notice that in second row, less engines are running than in fourth row, indicating the second row only uses electricity from the first panel.

Craft file

24-Aug-2014: verified for 0.24.2

Edited August 24, 2014 by Kasuha

[LordFjord]

rep++ for you, well deserved!

And the same thing seems to be valid for all resource-generating and consuming parts. Interesting.

[Kasuha]

Experiment: behavior of fuel tanks with disabled fuel crossfeed

Since there are no parts which are fuel tank and have disabled crossfeed at the same time in stock game, this experiment is made using modded KSP with fuel crossfeed disabled for the FL-T100 fuel tank.

Experiment demonstrates its behavior when fuel is both enabled and disabled in it, and in all kinds of attachment to other fuel tanks, i.e. through fuel pipe, radially as child, or axially.

Tanks with fuel disabled behave exactly like normal parts without fuel crossfeed.

Tanks with fuel enabled allow drawing fuel from parts attached through fuel pipe and from themselves, but not through axial or child radial connection.

Craft file

Javascript is disabled. View full album

[Red Iron Crown]

Kasuha, have you done any testing in 0.24.x? I saw a report today of fuel flowing into a tank from a radially mounted parent here.

[Padishar]

This hasn't changed recently. Post #8 describes this effect due to the tank being disabled.

[Red Iron Crown]

Ah, I didn't notice that it had been disabled in the linked post, thanks Padishar.

[Red Iron Crown]

Something funny is going on with LFO flow to the Vernor thruster. It doesn't seem to need fuel lines or crossfeeding parts to connect to a fuel source, which made me think it works like monoprop. But unlike monoprop, Vernors on a stage that has run out of fuel will not fire, even if there is fuel elsewhere on the craft like a monoprop RCS thruster would. It seems to follow rules different from both.

Edit: It appears that this behavior only occurred in my 0.24 game; updating to 0.24.2 made it behave like monopropellant as noted in the experiments below.

Edited August 20, 2014 by Red Iron Crown

[Padishar]

Something funny is going on with LFO flow to the Vernor thruster. It doesn't seem to need fuel lines or crossfeeding parts to connect to a fuel source, which made me think it works like monoprop. But unlike monoprop, Vernors on a stage that has run out of fuel will not fire, even if there is fuel elsewhere on the craft like a monoprop RCS thruster would. It seems to follow rules different from both.

https://dl.dropboxusercontent.com/u/61004449/ksp/RCSFlow.png

Have you tried with a different one of the fuel tanks empty to confirm that it is definitely the ones in the same stage that don't work?

[Kasuha]

Experiment: Vernor engines draw liquid fuel and oxidizer according to rules for monopropellant

Command pod is root part. Although all engines are active, fuel is drawn from below the decoupler first and above the decoupler second by all engines. That corresponds to monopropellant draw rules.

[Craft file]

Javascript is disabled. View full album

Edited August 5, 2014 by Kasuha

[Kasuha]

Experiment: O-10 engines draw fuel according to rules for monopropellant

Command pod is root part.

O-10 engines appear to draw fuel (monopropellant) independent on where they are mounted, using tanks behind a decoupler first. That corresponds to rules for monopropellant.

[Craft file]

Javascript is disabled. View full album

[Mesons]

The detail in this thread is incredibly informative and helpful! Thank you very much for taking the time to experiment and suss out all this data!

[undercoveryankee]

Experiment: Vernor engines draw liquid fuel and oxidizer according to rules for monopropellant

Command pod is root part. Although all engines are active, fuel is drawn from below the decoupler first and above the decoupler second by all engines. That corresponds to monopropellant draw rules.

If you look at the Vernor's config file, ModuleRCS accepts a resourceFlowMode value that apparently overrides the mode set by the fuel's definition. The Vernor is indeed configured to draw fuel using the STAGE_PRIORITY_FLOW mode that monopropellant uses.

MODULE
{
        name = ModuleRCS
        thrusterTransformName = RCSthruster
        thrusterPower = 12
        resourceName = LiquidFuel
        resourceFlowMode = STAGE_PRIORITY_FLOW
        PROPELLANT
        {
                name = LiquidFuel
        ratio = 0.9
                DrawGauge = True
                resourceFlowMode = STAGE_PRIORITY_FLOW
        }
        PROPELLANT
        {
                name = Oxidizer
                ratio = 1.1
                resourceFlowMode = STAGE_PRIORITY_FLOW
        }

Currently, all of the stock RCS parts have their fuel flow modes defined on the part. None of the engines that operate on engine controls have local flow modes set, so all of those engines (both O-10 and conventional LFO engines) draw fuel based on the fuel's settings.

[AlexanderB]

Hmm, I found something weird that might need some checking out. I think alt-click copy-pasting of parts does something odd with air intakes..

Gonna test a bit more, brb.

[Kasuha]

Hmm, I found something weird that might need some checking out. I think alt-click copy-pasting of parts does something odd with air intakes..

Gonna test a bit more, brb.

I agree with you, I had similar experience. Sometimes the order of part placement in the copied part changes.

What I have found, though, is that it is safe to drag and drop the assembly into the subassemblies list (destroying it in the process, i.e. dragging the original there, not the copy) and then you can copy it any number of times you wish as a subassembly and it is in correct order every time.

[AlexanderB]

In my case: make airplane with bi-coupler at the back for two engines, build wings with symmetry, etc.

-add 4 radial intakes on the wings (2x2 w/ symmetry)

-add 1 engine

-add 4 radial intakes on the wings (2x2 w/ symmetry)

-add 1 engine

(I probably used alt-click to copy-paste the intakes, I use it a lot to build quicker, and it would explain the weirdness.)

Caused the last-placed engine to flame out at 20k, while the fuel bar was still showing 0.04 intake air?

[Kasuha]

In my case: make airplane with bi-coupler at the back for two engines, build wings with symmetry, etc.

-add 4 radial intakes on the wings (2x2 w/ symmetry)

-add 1 engine

-add 4 radial intakes on the wings (2x2 w/ symmetry)

-add 1 engine

(I probably used alt-click to copy-paste the intakes, I use it a lot to build quicker, and it would explain the weirdness.)

Caused the last-placed engine to flame out at 20k, while the fuel bar was still showing 0.04 intake air?

Seems strange to me, nothing seems to have changed on that since previous release, I already built some planes and it all worked.

[Kasuha]

I have run all experiments with craft file under 0.24.2 and nothing has changed on them. So I assume these rules still apply on 0.24.2.

[AlexanderB]

I "solved" it for now by putting the jets on the wings, in case of asymmetric flameout I can just revert to SPH, take off and re-place the wing-assemblies and it should be ok.

[ratchet freak]

about the intake air order what did the intake air in the resource panel look like?

was there any intake air in the third case where the distribution looped over?

[Kasuha]

about the intake air order what did the intake air in the resource panel look like?

was there any intake air in the third case where the distribution looped over?

Good question. Yes, there was. Apparently the loop around happens through transfer from previous frame.

Edited September 2, 2014 by Kasuha