TAnks are emptying weird

[CosmicCharlie]

The stacks of 3 aren't emptying right. Some are emptying from furthest to nearest, but at least one is emptying from the top tank and bottom tank. Makes it hard to steer.

Anybody got an idea?

 

[Snark]

Holy mackerel, what a contraption. I gotta ask, why...?

Hard to tell from the picture. What's your fuel duct setup?

[Malich]

As Snark said... in that it could be anything. At  guess its some artefact of the design. I would suspect somewhere in that pile of tanks you have some parts with crossfeed enabled that is allowing them to feed fuel so the tanks empty top to bottom s per the KSP standard. In other I suspect is is disabled which means it's only feeding from the first tank in the stack. 

With a combination of these you'll have some stacks where it's feeding top to bottom, others where it's only feeding from the bottom tanks and some where, just to be awkward, the engine is using fuel directly from the bottom tank but the top tank is being depleted to refill a tank lower down the stack somewhere.

What re you lunching that needs  lifter like that? it is possible to launch  3-400 ton payload for a much less complex design that the one you've got there?

Edit - Remember, fuel ducts feed in the direction they were attached, so from the first tank you click on to the second, not specifically outside in or top to bottom.

Edited January 18, 2016 by Malich

[CosmicCharlie]

I don't have any fuel ducts. And I didn't change any crossfeed settings. On purpose anyway.

It's my Duna-and-back lander. 

Or it was. It's morphed. Now I'm just trying to get as much fuel as possible into orbit for refueling purposes.

But. It's got about 10,000 m/s delta-v. Landing on Duna is 6500. Plus I don't know how much getting back. I do have plenty(!) of fuel left over at the end, I admit. But what would I do differently?

 

[Edit] Well.  One stack of tanks isn't emptying at all. But the engine is firing. No fuel ducts.

It might have started when I added reaction wheels. But wouldn't they affect every stack the same way, ffs?

 

[More edit] Just took it apart and put it back together. Same thing. Maybe I'll try fuel ducts...Tried fuel ducts. Same thing.

See how 2 of the stacks are a lot emptier than the others? One of the stacks of two has both tanks emptying at the same rate. And they all have varying amounts of fuel left. How?

Edited January 18, 2016 by CosmicCharlie

[WhiteKnuckle]

I'd start by taking all the reaction wheels out and moving them to the top of the stacks underneath the nosecones. You say the problem started when you added them, and unless things have changed significantly since 1.0, the large SAS units were not very structurally sound, and would often cause rockets to break in half when you put one between heavy parts.

 

My guess is that one or more of the stacks is actually breaking in half at the SAS unit, if everything is strutted up your rocket might not be flying apart, but the fuel feed might be broken.  

[Malich]

A quick click of F3 would identify any failures if that was the case. 

[CosmicCharlie]

No failures in F3. And it does really screwy things. Like one stack of tanks is not emptying at all but the engine is still burning. With no fuel ducts.

 

Here's another datum: This basic design worked at least once. Then I emptied the inner tanks and saved under a different name - worked fine. Then I took that save and filled the tanks back up - problems.

 

Could that have anything to do with it?

 

[Edit] Moved the reaction wheels up under the nose cones. Fixed it. How, I'd like to know. How could they produce different results with different stacks of tanks?

Edited January 18, 2016 by CosmicCharlie

[Snark]

By the way-- if you're not using fuel ducts, you really should.  If you set it up properly, you can make it so that the outermost tanks drain completely, while all the other tanks stay completely full:  essentially, those outer tanks are fueling all the engines.

That way, as soon as those outer tanks are drained, you just activate decouplers and cast them off; they're dead weight you don't need.  Then the next-outermost tanks start to drain, and so forth.

In other words:  Instead of draining all tanks at once, you just drain them two at a time and throw away the empties as you go.  You'll end up with a lot more fuel left over when you get to orbit than if you just burn everything in parallel.

 

Also:  I'm curious, why did you have the reaction wheels in the middle in the first place?  It will make the rocket much more wobbly, and is hideously horrible for aerodynamics.  Putting them up under the nosecones will result in a rocket that is much sturdier and more aerodynamic.

Edited January 18, 2016 by Snark

[CosmicCharlie]

2 hours ago, Snark said:

Also:  I'm curious, why did you have the reaction wheels in the middle in the first place?  It will make the rocket much more wobbly, and is hideously horrible for aerodynamics.  Putting them up under the nosecones will result in a rocket that is much sturdier and more aerodynamic.

I thought the nearer the center of mass, the better. I still think that (why not?), but I'll put them under the nosecones.

 

Thanks for explaining about the tanks. I knew about asparagus arrangements, but it didn't make sense.

 

[Edit] Ok, I still don't understand. I need those outer engines to last two tanks worth and not one, to drain the inner stacks of 3 tanks enough so that the inner engines can continue to accelerate the ship when the outer engines jettison. If I drain the top outer tank into the stack of three, I end up with acceleration down instead of up, because the engines are pushing three full tanks instead of one when I jettison the outer tanks.

I'm missing something. What is it?

Wait. Decouplers in between tanks? That still wouldn't fix my downward acceleration problem, because I'd still end up with three full tanks per engine. And if I decoupled a top tank, wouldn't it just fall into the lower tank?

I'm missing something.

Oh. Not connecting the tanks right. So, (from the diagram on the asparagus staging page) I jettison the S4 tanks first, then S3, etc? I still end up with three full tanks per engine.

And how do I connect the outer tanks to the inner ones? I don't just have S1, I have 8.

Heeelllllp!

Or just do it to the outer tanks?

My mind is blown.

Edited January 18, 2016 by CosmicCharlie

[Snark]

52 minutes ago, CosmicCharlie said:

I thought the nearer the center of mass, the better.

Nope.  Reaction wheels have exactly the same effect (in terms of torque for turning the ship) regardless of where they're placed.

53 minutes ago, CosmicCharlie said:

I still think that (why not?)

Because physics! 

Torque is torque.  Doesn't matter where it's applied.  (I'll save the lengthy lecture in case you're interested, but unless you want to hear a bunch of physics talk, just trust me on this one.  Not only is this what the laws of physics say, but it's been experimentally verified in KSP as well.)

So, to sum up:  putting reaction wheels in the middle like that is bad, because,

• it will make the ship structurally wobbly
• it will ruin aerodynamics
• it won't actually accomplish anything, at all, in terms of trying to turn the ship.

[WhiteKnuckle]

53 minutes ago, Snark said:

Nope.  Reaction wheels have exactly the same effect (in terms of torque for turning the ship) regardless of where they're placed.

Because physics! 

Torque is torque.  Doesn't matter where it's applied.  (I'll save the lengthy lecture in case you're interested, but unless you want to hear a bunch of physics talk, just trust me on this one.  Not only is this what the laws of physics say, but it's been experimentally verified in KSP as well.)

 

Now hold on a sec. I know reaction wheel placement doesn't matter in KSP, but shouldn't it in real life?

 

Maybe I'm missing something basic here, but shouldn't increasing the moment from the CoM make a reaction wheel more effective?

[Padishar]

5 hours ago, CosmicCharlie said:

[Edit] Moved the reaction wheels up under the nose cones. Fixed it. How, I'd like to know. How could they produce different results with different stacks of tanks?

Do you have the version of the craft with the reaction wheels in the middle that behaves oddly saved?  Can you upload it somewhere?  I'd really like to understand what is happening here as they really shouldn't affect the fuel flow...

[CosmicCharlie]

Did asparagus staging.  Guess what. Differential emptying of tanks that should be emptying at the same rate.

 

 

Edited January 18, 2016 by CosmicCharlie

[CosmicCharlie]

48 minutes ago, Padishar said:

Do you have the version of the craft with the reaction wheels in the middle that behaves oddly saved?  Can you upload it somewhere?  I'd really like to understand what is happening here as they really shouldn't affect the fuel flow...

Yeah. Where do I upload it?

Edited January 18, 2016 by CosmicCharlie

[Padishar]

Any kind of file sharing service should do, dropbox, google drive, hastebin etc.  Then post the link to it here...

[FancyMouse]

3 hours ago, WhiteKnuckle said:

Now hold on a sec. I know reaction wheel placement doesn't matter in KSP, but shouldn't it in real life?

 

Maybe I'm missing something basic here, but shouldn't increasing the moment from the CoM make a reaction wheel more effective?

Nope in general if it's a closed system (including reaction wheel, excluding RCS), the torque it provides has nothing to do with where it is.

Imagine this - if you have a pair of force, of magnitude F each, in opposite directions, exerted d distance apart in perpendicular direction. You'll have F*d torque regardless of where this pair is put w.r.t. CoM, because if you calculate from CoM, the distance to CoM will be cancelled between the two forces, and only the distance difference d matters.

What KSP doesn't make sense is that, reaction wheel can provide constant torque regardless of current state of the wheel - that isn't right. To achieve that, real life reaction wheel would have to accelerate the wheel constantly unlimitedly, which isn't possible.

[Snark]

3 hours ago, WhiteKnuckle said:

Maybe I'm missing something basic here, but shouldn't increasing the moment from the CoM make a reaction wheel more effective?

But you're not "increasing the moment".  Torque is torque.

You're familiar with this equation for the force needed to get a certain acceleration, yes?

F = ma
("force = mass times acceleration")

The rotational equivalent is:

τ = Iα
("torque = moment of inertia times angular acceleration")

You'll note that the latter formula doesn't have any term whatsoever for where the torque is applied.  Doesn't matter.

Lots of people get this mixed up.  It's confusing because it seems counter to our physical intuition.  The problem is that our physical intuition screws up when it's presented with a situation that's outside our experience (i.e. a free-floating ship in space, applying internal torque); all the situations where we experience torque in daily life generally involve forcing the center of rotation to be the same place as where the torque is:  wheels on axles, a hand holding a hammer and waggling it, that sort of thing.  And it's true, changing the center of rotation matters... but not because you've moved the torque, it's because the moment of inertia of the object changes when you move the center of rotation.

A free-floating ship in space using its reaction wheels to rotate will always rotate around its center of mass, regardless of where the reaction wheels are placed.   Not just in KSP, but in real life too.  It is physically impossible to move your center of mass without either applying an external force (such as gravity) or throwing mass overboard (such as a rocket engine).  I think this is what throws people off-- they expect moving the reaction wheel to somehow change the center of rotation.

[CosmicCharlie]

7 hours ago, Padishar said:

Any kind of file sharing service should do, dropbox, google drive, hastebin etc.  Then post the link to it here...

https://drive.google.com/file/d/0B6Dffk6hHIF3d2pZb3g2SjE0WDQ/view

Let me know if you figure it out.

 

And if you have all the time in the world, you can try to figure out why this asparagus arrangement isn't working right. I've got the fuel ducts right (I think), but the tanks that should be emptying at the same rate, aren't.

https://drive.google.com/file/d/0B6Dffk6hHIF3N3JKejJ0UGhMZW8/view?usp=sharing

So then when I went ahead and waited til it was time to jettison the first two stacks, only one decoupled. 

Edited January 19, 2016 by CosmicCharlie

[Snark]

6 hours ago, CosmicCharlie said:

[Edit] Ok, I still don't understand.
...
Heeelllllp!
...
My mind is blown.

Heh, no worries, we were all there once. 

It sounds like you're basically asking "how does asparagus work" or "why does it help."  Yes?

So here's the deal:  When you're launching a rocket, every ounce counts.  You really need the dV, every bit of mass is counting against you.  So rockets need to be lean and mean.  There shouldn't be anything on the rocket that doesn't have to be there.  No dead weight allowed.  And anything that's initially necessary, but later becomes not necessary, we throw away as soon as we don't need it anymore because at that point it's dead weight and dead weight is bad.

It's the basic principle of why we have multi-stage rockets in the first place.

Right?  I'm willing to assume we're still on the same page here. 

Okay, so:  asparagus staging.  It's simply carrying the idea to its logical conclusion.  It's addressing two issues:

• Any engine that is on your rocket, and isn't firing, is dead weight.  So ideally, you'd like every single engine to be firing all the time.
• Any empty fuel tank space is dead weight.  So ideally, you want it to be the case that at all times during flight, the amount of empty fuel tank space is at a minimum.

Long-winded explanation of asparagus staging here.
 

Spoiler

So let's start with a simple, one-stage rocket.  You've got a big fuel tank carrying 32 tons of fuel, and there's an engine underneath.  We'll assume that you've designed it "correctly" so that you have the right size engine on it for that mass.  Too small is bad, because the engine's not powerful enough and you have a crappy TWR and too much gravity losses.  Too big is also bad, because then you're wasting mass by carrying more engine than you need.  So we'll assume that you have exactly the right size engine to launch it.

Got it pictured?

Okay, now consider how it flies.  It starts at 100% full, and burns until it's 0% full.  So on average, half of your total fuel tank space is empty for the duration of the burn.  Not so great.

Now, let's say we want to scale up our operation.  We want to launch with 64 tons of fuel, not just 32.  And we'll also need twice as much engine to launch it.  Well, we just double it, right?  For example, we could take the rocket from our first example, and put a ring of radial liquid-fueled boosters.  Let's say there's 8 of them, with 4 tons of fuel each.  And each one has a little engine on it that's 1/8th the power of the big central engine.

We just doubled our mass, and doubled our engine power, so we have the same TWR.  (Let's assume that the little engines have the same fuel efficiency and proportional power as the central one, they're just smaller.)  If we just launch with all engines firing at once, and they all burn together draining their tanks together, then this ship will perform identically to our first example.  We'll get the same overall efficiency, and end up with the same percentage payload to orbit.  We just doubled everything, that's all.

BUT!  We can be smarter about it.  We set up fuel ducts so that successive pairs of the radial boosters drain inwards to the central core.  (Like the diagram on the asparagus staging wiki, just with 8 radial tanks instead of 6).  You take off on all 9 engines (big central core, and all the radial boosters), but only two of the tanks are emptying.  Instead of every single tank emptying together, you have 7 tanks that aren't emptying at all, and 2 tanks that are emptying really fast.

And here's the brilliant bit.  As soon as those two tanks are empty, you jettison them and their engines.

You don't need the tanks:  they're empty, they're dead weight.  You get rid of them.

You also don't need the engines:  the rocket is lighter now than it was at takeoff, it doesn't need as much rocket power, so you ditch them.

And then the next pair runs out, and you ditch them; and so forth.

By doing this, every engine is burning all the time, and also you're running at an average of nearly-full fuel tanks.  It's much more efficient and works great.

Now, the downside of asparagus staging over doing a single very tall skinny rocket is that aerodynamics matters, and those radial boosters have drag, so you need to balance how things work.  In general, tall skinny designs are more aerodynamic.  But there are times when you need to launch something massive, and doing a single tall stack won't work, either because it would be so tall it would flop like a noodle, or because it simply wouldn't have enough thrust because you can't put enough engine power under one stack.

So to get back to your ship:  You're basically just launching a bunch of flying fuel tanks, right?  i.e. you have no real payload, other than the fuel itself?

So you want to arrange it so that you aren't carrying empty fuel tankage.  It's dead weight.  And also, as you burn fuel, your rocket gets lighter, which means you don't need as much engine power as you did on the launch pad, so you want to gradually ditch the engines as you lose mass, too.

It's hard for me to tell the exact arrangement of all your tanks and engines because you've got such a jumble there.  But basically, what you want to do is to arrange it with fuel ducts so that all tanks "drain inwards":  that is, an inner tank only starts using any fuel when all the tanks outwards from it are empty.  So the outermost stacks drain first while everything else stays at 100% full fuel load, and as soon as a stack is empty, you jettison the stack.

[Plusck]

22 hours ago, CosmicCharlie said:

See how 2 of the stacks are a lot emptier than the others? One of the stacks of two has both tanks emptying at the same rate. And they all have varying amounts of fuel left. How?

I tried your craft file. Unfortunately I don't have MechJeb so I couldn't use it as is, and manually deleting the MechJeb part from the file probably messed up the build.

However, assuming that deleting the MechJeb part is not partly the cause of the problem...:

- with the reaction wheels moved, the craft doesn't appear to be constructed properly. Some of the radial tanks have proper radial symmmetry (so adding a fuel line with 2x symmetry leading from one to the next works on both sides of the craft), but two of them don't (so trying to add a fuel line with 2x symmetry puts a fuel line on either side of the tank, not either side of the vessel).

- you shouldn't need fuel lines leading from the top to the bottom tank on the radial boosters. If you need them, the vessel has a problem and the tanks aren't constructed properly. If you don't need them, then adding them is going to complicate the fuel-finding algorithm and could well be causing unexpected results.

So I would suggest you remove all of the radial tanks, then build one again with 2x symmetry, take a copy of it with Alt-click and copy it over another three times around the vessel, then add fuel lines with 2x symmetry.

[CosmicCharlie]

Thanks I'll do it.

[WhiteKnuckle]

Thank you FancyMouse and especially to Snark for showing me where the fault in my thinking was. Since my edumaction in physics is entirely of the backyard variety  (automotive and aviation) it was easy to miss the very important point that the reaction wheel doesn't act on anything external, and therefore are unlike just about anything we deal with on a daily basis, even though the application of torque is something we're experiencing almost constantly.

 

[Plusck]

I took the liberty of playing around with that craft file.

Since everything was built with 8x symmetry, it was necessary to take it back to the core. So as floating items I kept one of the outer sets and one of the inner sets, then placed them with 2x symmetry, then alt-clicked the inner girder thingy and placed another three radially with 2x symmetry, then added fuel lines asparagus-syle going all the way back to the core. Since it was built with 2x symmetry, it was also possible to get rid of a number of the reaction wheels.

KER told me it had 12k m/s. Craft file: REFUEL reworked but without the MechJeb part.

Spoiler

The only problem with the staging is that you need to kill the engines when staging the inner circle of boosters because they are too close and may cause damage. So basically throttle off, stage, throttle up very slowly for a couple of seconds then full throttle again.

I got to orbit with 8800 m/s remaining in the tanks.  http://i.imgur.com/9rUzkWh.png