Fuel Line Woes!

[Paranoid Shark]

I have been trying for ages to create a VSSTO that has a mining kit onboard that will be able to return from everywhere but eve based on a clipped Mk3 fuselage. I'm pretty sure I have the design finalised, but I cannot work out the fuel routing. If anyone is able to help me out it would be greatly appreciated!

I have tested it out in orbit and it handles well up there and deorbits perfectly, it is just getting there that is the problem! In all my attempts the fuel either draws unevenly from one side tank and not the other if I try to set the fuel lines so that all the engines attached to the Mk2 Bicouplers drain from the bottom up. If I manually move fuel around and do a good gravity turn, it will get to at least a circular 80km orbit, which is perfect for rendezvousing with my fuel dock.

If I let it drain from the top down it is really unstable!

Here is the album

Javascript is disabled. View full album

and the google drive link

https://drive.google.com/folderview?id=0B7yXK3T8ki5gQndOVmZyVnRqQnM&usp=sharing

Thanks for your help,

Kit

[Snark]

My understanding of what you're trying to do is, you want to have a set of engines that drains from the bottom up. Is that correct?

Assuming that to be the case, here's the deal:

Engines drain from the "most distant" tank to the "nearest" tank. That's built-in, and there's no way to stop that. If you have a stack of fuel tanks, and the engine is on the bottom of the stack, then it will drain from the top down and there's no way around that with fuel lines, because stack-mounted tanks take priority over fuel lines and you'll badly confuse it if you try.

The only way I can see to make it drain a stack from the bottom up is to make sure that the engines are not part of the stack. For example, consider this:

The fuel will drain in the direction of the arrows, so the tanks will get used in the order A, B, C, D, E. That's because the engines aren't actually attached to the central stack, and their only connection is via the fuel lines. Therefore, as far as the engines are concerned, A is the most distant tank and gets drained first, followed by B, etc.

Does this help?

[Paranoid Shark]

Yeah, that is exactly what I want to happen, however I need to have some engines on the bottom tank as well to get my TWR high enough. I am happy for the two spikes that I have on the bottom to only draw from the main stack and to burn out early (the TWR is above 3 when they do) Your picture helps, but I have 4 "E" tanks (the Bicouplers) but only two of the bicouplers are joined with fuel lines to the main stack (Asthetical and due to lengths of fuel lines). The other two are attached to the tanks that are in your "E" Position, again burning a little bit longer than the engines in the "E" position. Could the problem be that the ISRU is attached to the node on top of one of the bicouplers?

Thanks,

Kit

[Snark]

Yeah, that is exactly what I want to happen, however I need to have some engines on the bottom tank as well to get my TWR high enough. I am happy for the two spikes that I have on the bottom to only draw from the main stack and to burn out early (the TWR is above 3 when they do) Your picture helps, but I have 4 "E" tanks (the Bicouplers) but only two of the bicouplers are joined with fuel lines to the main stack (Asthetical and due to lengths of fuel lines). The other two are attached to the tanks that are in your "E" Position, again burning a little bit longer than the engines in the "E" position. Could the problem be that the ISRU is attached to the node on top of one of the bicouplers?

If you need something under the central stack in the above setup, that can be problematic. If you mount it directly to the tank stack, it'll drain from the top down. You could prevent it from draining the stack by putting something in between it and the stack which isn't crossfeed-capable (like a stack decoupler)... but then there's the problem of how to supply that engine with fuel. You might be tempted to try to run fuel lines from the "E" tanks down to it, but that way lies madness. The problem is that KSP wants fuel flow to be a tree. It gets confused if you have any loops. For example, it's fine to have two radial engines pulling from a central tank, or one central engine pulling from two radial tanks ... but it's not okay to have two radial tanks pulling from something in the middle, and then have something in the middle that pulls from the two radial tanks. If you do that, you're likely to get lopsided fuel flow, which is a problem.

About the only way I could see it working would be if you don't add one thing under the central tank, but two (or, at least, as many as you have "E" tanks). For example, in the diagram above, imagine putting two engines (call them X and Y) under the central tank, with a stack decoupler above them to stop them from draining the stack. Engine X would be slightly to the left, and would have a fuel line running from the left "E" tank to it. Engine Y would be slightly to the right, and would have a fuel line running from the right "E" tank to it. Note that you could not put X and Y on a single bicoupler tank and run two fuel lines from the two E tanks to it, since that would run into the "fuel loop" problem I described above. You have to make sure that your fuel flow is a tree, not a DAG.

I doubt that the ISRU is an issue.

Edited November 3, 2015 by Snark

[Highguard]

It's quite possible to draw fuel bottom-to-top in a single stack, with engine at the bottom but there are two 'issues'.

The first is; how to run fuel lines from the top tank to the bottom one, since they'll just clip into the first surface they meet.

For that, attach a couple of cubic octagonal struts in place of the fuel tanks marked 'E' in Snark's handy picture. Run the fuel lines from the top tank to the top of struts, more or less as illustrated.

Then run another pair of fuel lines from the bottom of the struts directly to the engine itself. The engine 'sees' the fuel lines and takes fuel from the top tank 'D' (via the struts), which pulls fuel from 'C', which drains 'B', which refills from 'A' - so the tanks empty from bottom to top.

The second issue is when you can't access the engine for some reason (blocking structural parts or something).

In that case you can attach the struts to a higher tank (eg; 'B') and run the fuel lines 'D' -> struts -> 'A'.

This gives a slightly odd fuel-consumption pattern since 'A' is kept full. Nevertheless, the other tanks empty in the order 'B', 'C', 'D' which is probably sufficient for most purposes.

[BoilingOil]

When you have your central engine decoupled under the main stack, why not let it share fuel from the side engine's tanks (tanks E)? You *could* use fuel lines for that!

[Snark]

When you have your central engine decoupled under the main stack, why not let it share fuel from the side engine's tanks (tanks E)? You *could* use fuel lines for that!

Because that creates a multi-path fuel setup (e.g. fuel could go from D -> E1 -> engine or D -> E2 -> engine), which KSP does not like. It will likely cause weird behavior such as draining one of the E tanks before the other one. Fuel flow needs to be a tree-- that is, there needs to be one and only one path from any particular tank to any particular engine. See my post up above.

[A_name]

I don't know enough about fuel line mechanics to help on that, but you may want to take a look at TAC Fuel Balancer. Also, you may want to lock the upper most tank and unlock it only after everything else is dry. This is more of a work around than a solution but it beats micromanaging the tanks all the way up as I understand you're currently doing.

[BoilingOil]

Because that creates a multi-path fuel setup (e.g. fuel could go from D -> E1 -> engine or D -> E2 -> engine), which KSP does not like. It will likely cause weird behavior such as draining one of the E tanks before the other one. Fuel flow needs to be a tree-- that is, there needs to be one and only one path from any particular tank to any particular engine. See my post up above.

You mean that BOTH lines ending up in the same engine would be the problem? Yeah, that would explain some things... Maybe some day I'll be an engineer, but I'm not one today, yet

[Paranoid Shark]

Sorted out the fuel "tree" and it is now draining the way that I want to. However when it gets low on fuel it has a tendency to do a loop when you are almost finished with the gravity turn. How can i stop this?

Thanks,

Kit

[Snark]

Sorted out the fuel "tree" and it is now draining the way that I want to. However when it gets low on fuel it has a tendency to do a loop when you are almost finished with the gravity turn. How can i stop this?

By "do a loop," I assume that you're talking about flight characteristics and not fuel flow, yes? i.e. that your gravity turn is working well, but then when low on fuel it has a tendency to "flip" the wrong way round?

What altitude does that happen at?

If it's way up high when the atmosphere gets really thin (e.g. over 20 or 25 km), just add reaction wheels for stability.

If it's lower down, then this is just the usual KSP problem with aerodynamic stability. You need your CoM to stay in front of your center of drag. If you were relying on fuel mass up front to keep your CoM forward, then it's not surprising that this could happen when fuel gets low. Either figure out some way to shift your CoM forward (that involves something other than fuel mass), or else add drag / stability to the back, such as adding fins or other control surfaces to the rear end of the craft.

[DancesWithSquirrels]

Set up your stack with three tanks, split them so they don't crossfeed, then fuel-duct the top and bottom into the center tank. Run lines from center tank your engine(s), and the ends will drain simultaneously. Here's an example:

I really REALLY want a tweakable in the editor to disable tank/tank feeds for this very reason. Or maybe a very low mass plate I could insert to block flow without changing the aerodynamics. Stack separators are heavy and end up in the staging whether I want them to or not.

EDIT:

Duh! Of course that works, because the stack drains through the middle. This is another way to set it up for evenly balanced drain - a stack with the fuel line exiting in the middle will drain from both ends:

Edited November 3, 2015 by DancesWithSquirrels

[Paranoid Shark]

Thanks for all your help guys, adding a large reaction wheel to the bottom of the cargo bay sorted the flipping. Can leave MechJeb to it on ascent guidance and it will make it up to 80 x80 with about 50m/s to spare.

I think this is going to be a way better design when the small size ISRU comes out in 1.05!

[A_name]

Thanks for all your help guys, adding a large reaction wheel to the bottom of the cargo bay sorted the flipping. Can leave MechJeb to it on ascent guidance and it will make it up to 80 x80 with about 50m/s to spare.

I think this is going to be a way better design when the small size ISRU comes out in 1.05!

I'm also waiting on the small ISRU!

[Highguard]

Because that creates a multi-path fuel setup (e.g. fuel could go from D -> E1 -> engine or D -> E2 -> engine), which KSP does not like...

As explained in my previous post, KSP is perfectly happy with multiple fuel paths and will always drain source tanks evenly. What it doesn't like - and never has - is loops.