Piping issue or bug

[bzgzd]

I am not sure if this is some known bug or why is it like this...
I know there are problems when connecting machines outputs directly to "main" pipe but there seems to be problem also when splitting pipes.

Here all valves are set to 1kg and there are 10 of them. It could be machines needing 1kg water on input.

Left version is how I am connecting my machines normally but it seems to not work properly.
Right version is delivering water to all valves and pump is running 100% but on the left pump runs maybe just 50% time.

To me it seems like pipe junctions (splits) are not working very well. Connecting all water consumers to one line is not very good workaround when I have them all over my base.
Is this also some known and for now ignored thing?

 

[GuyPerfect]

Could you describe exactly the behavior that looks odd to you? Off the top of my head I can think of several scenarios where a setup like this can behave in unexpected ways just because of how pipe junctions work...

Whenever a pipe delivering a resource splits off in another direction, the junction attempts to distribute liquid/gas packets evenly by cycling through the connected pipes. For example, in a T-shaped junction, with water flowing from the bottom, the junction will alternate left and right each packet, regardless of how much water is actually in that packet.

When you have a large number of interconnected junctions like you do in your picture, the states of each junction remembering which pipe gets the next packet could result in some seemingly anomalous behavior, like watching a number count up in binary.

[bzgzd]

7 minutes ago, GuyPerfect said:

Could you describe exactly the behavior that looks odd to you?

I would expect pump going at 100% in both cases. There is same 10kg consumption so...

8 minutes ago, GuyPerfect said:

Whenever a pipe delivering a resource splits off in another direction, the junction attempts to distribute liquid/gas packets evenly by cycling through the connected pipes. For example, in a T-shaped junction, with water flowing from the bottom, the junction will alternate left and right each packet, regardless of how much water is actually in that packet.

This seems as ok feature but what it does when one path from junction is full (or almost full) seems to be wrong.
Maybe when 10kg water packet comes to junction, should go to the left by "alternate rule" but there is just 100g empty on that left path and so it fills 100g and stays frozen on junction not moving further or something?

 

[Risu]

1 hour ago, bzgzd said:

I would expect pump going at 100% in both cases. There is same 10kg consumption so...

There would only be 1 kg consumption every time the valve is the recipient of the water packet.
It only sees the missing mass in pipe at valve input and is trying to replace it.
 

[bzgzd]

Simplified version:

This pump is not going 100%.

It is probably like that for ages but to me it still seems like bug...

[The Flying Fox]

I've believe I too noticed this issue at some point in my current game.  I'm not sure if it's always been like this or its a new thing, or if it's just by design.  I had a particular T-junction in my oxygen pipe that I noticed was backing up the line on occasion and, from looking at what the packets were doing, deduced that packets tend to linger for an extra second at the T-junction deciding how much of the packet goes down which direction.  This, of course, only happens when one or more legs of the T-junction are backed up.  When both legs are free, then the packets just split evenly into the two legs with no delay. 

 

I ended up coming up with this little number to make the split work cleanly with no packet delay.

Each valve is set to 500G/s, so half the capacity of the pipe.  Of course, it won't truly split the stream in-half due to the top valve being before the second, but it works well enough for my needs.

[bzgzd]

28 minutes ago, The Flying Fox said:

I've believe I too noticed this issue at some point in my current game.  I'm not sure if it's always been like this or its a new thing, or if it's just by design.

Well I made bug report then and will see, probably will end as "known issue" ...

 

[Hechicera]

One issue I've noticed with T's in fluid and gas in-game is not just packet metering but the dreaded "Which way does this pipe go?"  I wish the game had backflow/flapper valves we could add at come of the junctions to control flow direction. I sometimes use the pipe cross-over alone with no crossing piping since it does indicate direction and that often helps clear things up. But it doesn't prevent backflow, the meters do though. Having a flow meter and a cross-over on each piping run is laggy and space inefficient. The meters do seem to slow things down even when set wide open and used as faux flapper valves for flow control.