Liquid pipes and valves flow

[sheaker]

Dear All,

I can not fully understand flow behavior behind this setup. Could You, please, explain why I can not get fluid to flow on first and second setup but the flow occurs on the third setup?

1st setup:

2nd setup:

3rd setup:

 

Best regards,

Sheaker

[psusi]

It looks like you are trying to use a shutoff to send liquid either to the left tank, or overflow to the right tank, but I don't see any input.  Where is this liquid coming from?

[sheaker]

14 minutes ago, psusi said:

It looks like you are trying to use a shutoff to send liquid either to the left tank, or overflow to the right tank, but I don't see any input.  Where is this liquid coming from?

I am sorry I didn't mark it at the very beginning. Here is the input flow. Dark blue is aquatuner bypass circuit and light blue is aquatuner cooling circuit. There is no fluid in second and third pictures because I get them after I fix it.

[psusi]

4 minutes ago, sheaker said:

I am sorry I didn't mark it at the very beginning. Here is the input flow. Dark blue is aquatuner bypass circuit and light blue is aquatuner cooling circuit. There is no fluid in second and third pictures because I get them after I fix it.

The green section appears to be letting the output from the right tank flow back to the input of that tank.  I was talking about the pipes above the tanks, which appear to route a missing input into either the left or right tank.

[Dracian]

I think it is because of a conflict between entrances and exit. I may be wrong but it looks like you have not put all entrances on one side and all exits on another.

 

Basically, your system must be like that :
 

>----\              /------>
         \_____/
         /         \
>----/              \------->

While yours looks like that

>----\            /------<
        \_____/
        /         \
<----/            \------->

Which means liquid doesn't flow in the middle section.

I'm sorry if this is not clear enough, but English is not my native language and I don't know how to formulate it correctly.

[sheaker]

11 minutes ago, psusi said:

The green section appears to be letting the output from the right tank flow back to the input of that tank.  I was talking about the pipes above the tanks, which appear to route a missing input into either the left or right tank.

I am sorry again. This setup is to precool liquid pipse for rocket fueling. So it shares some part of the circuit with fuel line. But this shouldn't be the case since small and not relevant change makes the liquid flow.

 

9 minutes ago, Dracian said:

I think it is because of a conflict between entrances and exit. I may be wrong but it looks like you have not put all entrances on one side and all exits on another.

 

Basically, your system must be like that :
 

>----\              /------>
         \_____/
         /         \
>----/              \------->

While yours looks like that

>----\            /------<
        \_____/
        /         \
<----/            \------->

Which means liquid doesn't flow in the middle section.

I'm sorry if this is not clear enough, but English is not my native language and I don't know how to formulate it correctly.

Ok, I know what You mean and I created a simplified case in sandbox for setup. However it is now working exactly in the same way. Here is example:

OK, I agree that between valve input and liquid vent flow may occur in both direction. Why the liquid vent does not act like a sink? I mean I think all routs should end on liquid vents.

Now modified part. Still not working:

Since modification is pretty the same as in my setup I would expect the same behavior. Why it isn't?

 

Now I know I should use one liquid pipe bridge to force one way flow in red part liquid line on the first image but I still don't get why flow occurred on third image.

[Olleus]

When you have loops without inputs/outputs the resolution of flow direction is essentially undefined. Now the source code of the game is deterministic so it will compute one particular direction. But it does so "accidentally" rather than by following the fundamental rules of pipe flow (ie: input takes priority over pipeflow, pipeflow takes priority over output, ect...) and so is unreliable. In technical language it would be described as an implementation detail rather than a design choice. In practice, it means that if a pipe network can't be decomposed into simple sections each that go from a bunch of green arrows to a bunch of white arrows, you should assume it won't do what you want it to.

 

So, to take your simple example. You have one stretch of pipe which is 2 green arrows from the pumps to the white arrow of the valve. This flows as normal. The remaining section has, the green/white arrow of the valve at either end and the white arrow of the liquid output in the middle. So the flow will be from green to white to white. In your second example you have a loop with a single white arrow and so, essentially, the whole thing is undefined and should not be relied upon to do anything sensible.

[psusi]

Insert a bridge between the shutoff and the vent.  The liquid won't flow past the shutoff because there is a source further down that line ( the other side of the shutoff ).

[TheDeamon]

5 hours ago, sheaker said:

I am sorry I didn't mark it at the very beginning. Here is the input flow. Dark blue is aquatuner bypass circuit and light blue is aquatuner cooling circuit. There is no fluid in second and third pictures because I get them after I fix it.

 

2 hours ago, psusi said:

Insert a bridge between the shutoff and the vent.  The liquid won't flow past the shutoff because there is a source further down that line ( the other side of the shutoff ).

Pretty much this, you're not getting liquid flow through the green area from the blue area because the liquid bridge in the top-left of the SS is "the direction of flow" that the game has decided to abide by. Inserting a liquid bridge after the second shutoff but before the input for the reservoir should allow liquids to flow as you seem to desire? (overflow returns to the reservoir)

[wachunga]

Trace a path from the bridge at the very top of the original images. It goes output -> input -> input -> output/output. You can't do that.  Similarly trace a path from the pumps in your other images. It goes output/output -> input -> input -> output. You've again confused the pipes.

There must be a clear one way flow from outputs to inputs. Outputs on one side and inputs on the other. That the third image "worked" is a quirk and should not be taken as a sign of doing things properly.

[TheDeamon]

26 minutes ago, wachunga said:

Trace a path from the bridge at the very top of the original images. It goes output -> input -> input -> output/output. You can't do that.  Similarly trace a path from the pumps in your other images. It goes output/output -> input -> input -> output. You've again confused the pipes.

There must be a clear one way flow from outputs to inputs. Outputs on one side and inputs on the other. That the third image "worked" is a quirk and should not be taken as a sign of doing things properly.

The third image worked because it confused the game engine just enough to allow for the overflow of the input to become an additional source for the very same input. It also created a "T" intersection in the pipes which then made that two pipe section of loop its own separate flow so far as the game engine could determine.

Basically his green line is two different "flows" so far as the game is concerned. One is the Bridge to the Reservoir input to the "T" which then goes to the valve input.

Then the second on is the Valve overflow to the "T" which then returns back to the valve input.

Easiest way to ensure proper pipe flow is pretty much as you say, but I'm going to rephrase it:

Don't give the game engine a choice on direction of flow.

If you're going to insert another fluid/gas input into a line after a series of inputs, place a valve or bridge in between those inputs and that new source. This will prevent the game from trying to "backflow" to the previous inputs from the new input source, as the valve or bridge will not allow the backflow to happen.

edit: need to fix the image

 

[Gurgel]

With the flow calculation optimization some time ago, there are a few situations where flow should occur intuitively, but does not. 

This is one of the situations where that happens:

5 hours ago, Dracian said:

 

>----\            /------<
        \_____/
        /         \
<----/            \------->

 

The fix is to use bridges as "flow diodes", since the flow calculation stops at them as they have only one direction. The new algorithm is more convenient in a lot of situations and a lot faster, so I do not consider this a bug. And once you know about the effect, it is very easy to fix.

[TheDeamon]

Yes, in this fixed image the Green flow line runs counter to the yellow one, That's also why that section has a star next to it. That section is going to have "failure conditions"  where its behavior may change during the session, or between sessions depending on various factors. Which is why it is marked for flow in both directions.

The yellow section will try to flow back into where it came from(because the other side is supposed to be flow from the bridge), but if it is obstructed on that tick, it'll then look for where else it can go, at which point it would see the reservoir is available(path is not obstructed), and may try to go there... Assuming nothing from the Bridge is otherwise obstructing the path. (T intersections get a little more complicated in that it alternately can decide to do a 50/50 split until a path is blocked)

But the moment that reservoir becomes full, the bridge would assume dominance over that section of pipe and the only option would be to go through the valve input at that point.

[sheaker]

3 minutes ago, TheDeamon said:

Yes, in this fixed image the Green flow line runs counter to the yellow one, That's also why that section has a star next to it. That section is going to have "failure conditions"  where its behavior may change during the session, or between sessions depending on various factors. Which is why it is marked for flow in both directions.

The yellow section will try to flow back into where it came from(because the other side is supposed to be flow from the bridge), but if it is obstructed on that tick, it'll then look for where else it can go, at which point it would see the reservoir is available(path is not obstructed), and may try to go there... Assuming nothing from the Bridge is otherwise obstructing the path. (T intersections get a little more complicated in that it alternately can decide to do a 50/50 split until a path is blocked)

But the moment that reservoir becomes full, the bridge would assume dominance over that section of pipe and the only option would be to go through the valve input at that point.

Thank You all for Your time. Here is "the winner" that reveal to me what happened here.
Thank You!

[TheDeamon]

9 minutes ago, sheaker said:

Thank You all for Your time. Here is "the winner" that reveal to me what happened here.
Thank You!

I imagine you now understand what I meant about "don't give it a choice" in all of this?

[sheaker]

1 hour ago, TheDeamon said:

I imagine you now understand what I meant about "don't give it a choice" in all of this?

Yes, sure. I already added one way liquid pipe bridge. Its on my previous post. The trick I didn't understand earlier was this small circle near the bottom valve input. Once the game choose one way 'upwind' then the second output became available.