1. Update previous videos for 1.0: hell yeah 2. Videos on new powerplants: hell yeah 3. Random test videos: hell, yeah. These have been especially enlightening
From the looks of it the game simply doesn't account for physical size of the junction in head lift calculation, effectively giving you up to +/- 2 meters of free head lift depending on its orientation. The test results at 4:33 would make perfect sense if you imagine the sideways pipes offset vertically (in their entirety together with their outputs) as if they originated on the axis of the center pipe instead of junction flanges. That would shift the top pipe 2 meters down and the bottom pipe 2 meters up as far as head lift is concerned. Then the actual height offsets would become +1/0/-1 on the left setup and -2/0/+2 on the right, giving you unintuitive flow priorities that the test demonstrated as the result. I still wonder why the flow would split evenly between the mid and low output even though the pipes don't seem to be full, but maybe that's just how this particular test setup works, or maybe there's some more magic to it.
For test 6 and 7 the pipes are very empty. The packagers are all running faster than the un-packager. For the flow in the draining tests I would explain it as the following. The junction pushes fluid to all pipes with a lower fill percentage than the junction. Which means the 3 destination pipes receive the same amount. The pipes then control the flow. The pipe flow models may only keep track of the 2 ends of the pipe and assume a straight line between them. The pipe going uphill would want the flow to go back into the junction. The flat and down hill pipes don’t have a reason to flow back to the junction, so they flow to the packager. The packager is running faster than the incoming flow, so there is space in the integer fluid storage in the packager preventing back flow.
@@BitwiseAssembly That makes sense. I did some testing and my theory turned out to be correct: pipeline junction indeed doesn't add any head lift. This allows you, for example, to pump liquids upwards for free by chaining vertical junctions. It also finally explains how VIP junction works: the two pipe junctions collectively eat 4 meters of head lift from the upper pipe, effectively shifting it 4 meters below the lower pipe. Every junction on top of it gets shifted further 4 meters down, which in the end results in the whole setup prioritizing the inputs starting from the lowest one despite them being physically at the same level.
Actually scratch that, it's not that simple. It seems to be a bug somewhere else, junctions only trigger it under certain conditions, but are not at all guaranteed to always work. More research is needed.
This is awesome. The difference of priorities between mk1 and mk2 setups is a nice find. This is a bit new to me, so correct me if I'm wrong, but it seems we could conclude perfect priority junction cannot be made with mk2 pipes, but it can function if the fresh/recycled ratios are such the deviation doesn't matter, like fresh water demand being > 17% as demonstrated in test rig 9. Other thing I noticed about the older alumina setup, I understand the unpowered pump makes it so the fresh water can't fill the vertical buffer, and that per test rig 4/8 higher headlift pipe in a flat junction takes "some" priority, I assume the alumina setup only works because 120m3/min recycled water vs 80m3/min fresh water fits within that "some" priority ratio? It would be nice if we could figure out a formula for the ratio based on headlift difference.
I would agree that a perfect priority junction cannot be made with mk2 pipes and I agree with your observations. There definitely is a priority limit in the older setup. I currently only know what recycle rates I have gotten to work when using it. The one shown in the video is 60% recycled water. I will work a a rest rig to see if I can measure the limit. From an anxiety stand point I do prefer the older set up, since the waste return pipe is not full. The new setup is so simple and reliable, that I am getting use too the pipes all being full and happy.
Great video! I would like to ask for an additional test. If you repeat tests 5 and 9 but removing the bottom pipe, would the same total priorization happen to the middle input? I ask this because this is how the VIP junction is usually built, and there are many times it doesn't work.
Thank you. I expect it will strongly favor the on level connection. I am working on a follow up video where I made direct liquid recycling test stands and measure the different recycling methods. I have the VIP junction included in Mk1 and Mk2 pipes. With Mk1 pipes it has 100% prioritized the on level pipe. With Mk2 pipes it’s around 86% comes from the on level pipe. The video is taking longer then expected the some of the test setups have 2 stable solutions based on initial conditions, and I want to be sure the results are repeatable.
@@BitwiseAssembly That's amazing that you also check de different initial conditions. Keep it up! I must say I just implemented test 5 for my alumina production, just a vertical junction with no pumps nor valves, and it seems to work. Before, my recycled water pipes kept clogging up after many hours.
found very interesting ! tho I didnt understand completely. maybe a brief explainer in laymens terms, this felt like it assumes a moderate-advanced fore-knowledge of the topic
I may have to watch the old videos, but I agree. I see the results, and the info is there, but I would love a bit more of the unstated conclusions and premises out loud. I'll check out past videos for sure.
If I'm understanding correctly my intuition was right, use lower pipes to assign priority essentially. I set up a quick bauxite system when I first unlocked it and did that, I haven't seen any issues but I also haven't checked on it (my Aluminum Sheets in the Dimensional Depot haven't ran out so I think I did everything right)
I just stumbled upon your videos, and loving them. insert Jesse Pinkman meme I have over 300 hours in this game, but never knew about these priorities. Do gases exhibit this behavior too ? Maybe that is a video :)
Man, is that a doozy of a question. Gases don’t use head lift. so the feedback loop rotating the junction causes, might not have an effect on gases. If it does tho…. It would be useful to know for the end game gasses.
I may have trimmed a few too many words from the script. It’s hard to get a good balance and this was my first talking video. I guess I could go over the major points here. Very beginning of the video I talked about how there was no real changes in 1.0 vs update 8 when it came to pipes. Which the exception of the saving system impact on pipe networks. In the last pipe video I did there was table for each network with included running at 100% measurement and that inadvertently measured impacts from saving and loading, which explained a lot of the complaints the community was having maintaining full flow. The pipe flow system in attempted to simulate liquid momentum by averaging new calculated flow with the previous cycles calculated flow, this caused the sloshing in the pipes. In update 8 when the game paused to save the previous cycle data would effectively be lost and any wave action would fall flat. This helped some networks and hurt others so when 1.0 no longer paused the game to save some networks performed better and others did worse. This is something that is so in the weeds I did not think anyone would benefit from it so I glazed over it. Moving on to everything is a fluid buffer. A lot of the fluid volumes are hidden to the player, which for most players is fine. Explaining the why is the hard part, the fluid mechanics is broken into two parts head lift which is used to limit what hight a liquid reach and fill percentage. These two processes interact with each other. head lift propagates through the network only through pipe/pipeline components which are full (100% or more full). The pipe network runs using floating point values with a soft limit of 140% full, this full range make the network act like a rubber hose but seams to be a necessary evil to prevent losing fluid in the network from sudden changes in flow. This is why everything has two values in the video a programed value and the overfilled value. Each of the different components needs to have an internal volume, so the head lift can propagate through it. The pipeline Junction programming was not adjusted when they added the ability to rotate it. That is why everything gets weird with it is rotated, all of that testing was just showing how I came to that. The created several Reddit posts using this test stand in update 8, this was the first video where I included them. I don’t have access to the program so some aspects I can’t 100% confirm. I am a mechanical electrical control systems engineering and part of my IRL job has me reverse engineering control systems from the UI and IO. So take that into consideration. I don’t think including the word salad above would have made the video clearer.
1. Update previous videos for 1.0: hell yeah
2. Videos on new powerplants: hell yeah
3. Random test videos: hell, yeah. These have been especially enlightening
Your junctions testing setups are brilliant. Finally we know how the VIP junction works. Thank You so much :)
This video is as important now for reference as the fluids pdf guide
Awesome work man seriously. My only suggestion is to turn down some of the game sound so that your voice is clearer. Liked and subbed.
Thank you for the feedback.
I will make adjustments for the next video.
@@BitwiseAssembly it is not that obvious, but the game has a BUNCH of audio sliders, you just have click on the tiny expand icon :)
I know. I even recorded my voice in a separate channel, so I would have more control.
But I failed to have a someone else review it before posting.
@@BitwiseAssembly that is already more than most peoples first videos. you'll do even better next time :)
Each of these SF mechanics-test videos has been such a lovely surprise
From the looks of it the game simply doesn't account for physical size of the junction in head lift calculation, effectively giving you up to +/- 2 meters of free head lift depending on its orientation. The test results at 4:33 would make perfect sense if you imagine the sideways pipes offset vertically (in their entirety together with their outputs) as if they originated on the axis of the center pipe instead of junction flanges. That would shift the top pipe 2 meters down and the bottom pipe 2 meters up as far as head lift is concerned. Then the actual height offsets would become +1/0/-1 on the left setup and -2/0/+2 on the right, giving you unintuitive flow priorities that the test demonstrated as the result.
I still wonder why the flow would split evenly between the mid and low output even though the pipes don't seem to be full, but maybe that's just how this particular test setup works, or maybe there's some more magic to it.
For test 6 and 7 the pipes are very empty. The packagers are all running faster than the un-packager.
For the flow in the draining tests I would explain it as the following.
The junction pushes fluid to all pipes with a lower fill percentage than the junction.
Which means the 3 destination pipes receive the same amount. The pipes then control the flow.
The pipe flow models may only keep track of the 2 ends of the pipe and assume a straight line between them.
The pipe going uphill would want the flow to go back into the junction.
The flat and down hill pipes don’t have a reason to flow back to the junction, so they flow to the packager.
The packager is running faster than the incoming flow, so there is space in the integer fluid storage in the packager preventing back flow.
@@BitwiseAssembly That makes sense.
I did some testing and my theory turned out to be correct: pipeline junction indeed doesn't add any head lift. This allows you, for example, to pump liquids upwards for free by chaining vertical junctions.
It also finally explains how VIP junction works: the two pipe junctions collectively eat 4 meters of head lift from the upper pipe, effectively shifting it 4 meters below the lower pipe. Every junction on top of it gets shifted further 4 meters down, which in the end results in the whole setup prioritizing the inputs starting from the lowest one despite them being physically at the same level.
Actually scratch that, it's not that simple. It seems to be a bug somewhere else, junctions only trigger it under certain conditions, but are not at all guaranteed to always work. More research is needed.
This is awesome. The difference of priorities between mk1 and mk2 setups is a nice find.
This is a bit new to me, so correct me if I'm wrong, but it seems we could conclude perfect priority junction cannot be made with mk2 pipes, but it can function if the fresh/recycled ratios are such the deviation doesn't matter, like fresh water demand being > 17% as demonstrated in test rig 9.
Other thing I noticed about the older alumina setup, I understand the unpowered pump makes it so the fresh water can't fill the vertical buffer, and that per test rig 4/8 higher headlift pipe in a flat junction takes "some" priority, I assume the alumina setup only works because 120m3/min recycled water vs 80m3/min fresh water fits within that "some" priority ratio? It would be nice if we could figure out a formula for the ratio based on headlift difference.
I would agree that a perfect priority junction cannot be made with mk2 pipes and I agree with your observations.
There definitely is a priority limit in the older setup. I currently only know what recycle rates I have gotten to work when using it.
The one shown in the video is 60% recycled water.
I will work a a rest rig to see if I can measure the limit.
From an anxiety stand point I do prefer the older set up, since the waste return pipe is not full.
The new setup is so simple and reliable, that I am getting use too the pipes all being full and happy.
Great video! I would like to ask for an additional test. If you repeat tests 5 and 9 but removing the bottom pipe, would the same total priorization happen to the middle input? I ask this because this is how the VIP junction is usually built, and there are many times it doesn't work.
Thank you.
I expect it will strongly favor the on level connection.
I am working on a follow up video where I made direct liquid recycling test stands and measure the different recycling methods.
I have the VIP junction included in Mk1 and Mk2 pipes.
With Mk1 pipes it has 100% prioritized the on level pipe.
With Mk2 pipes it’s around 86% comes from the on level pipe.
The video is taking longer then expected the some of the test setups have 2 stable solutions based on initial conditions, and I want to be sure the results are repeatable.
@@BitwiseAssembly That's amazing that you also check de different initial conditions. Keep it up! I must say I just implemented test 5 for my alumina production, just a vertical junction with no pumps nor valves, and it seems to work. Before, my recycled water pipes kept clogging up after many hours.
Thank you much!
found very interesting ! tho I didnt understand completely. maybe a brief explainer in laymens terms, this felt like it assumes a moderate-advanced fore-knowledge of the topic
There were a lot of topics here that I talked about.
Is there a particular part you need additional explanation on?
I may have to watch the old videos, but I agree. I see the results, and the info is there, but I would love a bit more of the unstated conclusions and premises out loud. I'll check out past videos for sure.
If I'm understanding correctly my intuition was right, use lower pipes to assign priority essentially. I set up a quick bauxite system when I first unlocked it and did that, I haven't seen any issues but I also haven't checked on it (my Aluminum Sheets in the Dimensional Depot haven't ran out so I think I did everything right)
I just stumbled upon your videos, and loving them.
insert Jesse Pinkman meme
I have over 300 hours in this game, but never knew about these priorities.
Do gases exhibit this behavior too ? Maybe that is a video :)
Man, is that a doozy of a question.
Gases don’t use head lift.
so the feedback loop rotating the junction causes, might not have an effect on gases.
If it does tho…. It would be useful to know for the end game gasses.
Great video, lot's of work put in it. But.. i didn't understand a half of it. May I suggest an explainer for some more dumb audience? :)
I may have trimmed a few too many words from the script. It’s hard to get a good balance and this was my first talking video.
I guess I could go over the major points here.
Very beginning of the video I talked about how there was no real changes in 1.0 vs update 8 when it came to pipes. Which the exception of the saving system impact on pipe networks.
In the last pipe video I did there was table for each network with included running at 100% measurement and that inadvertently measured impacts from saving and loading, which explained a lot of the complaints the community was having maintaining full flow.
The pipe flow system in attempted to simulate liquid momentum by averaging new calculated flow with the previous cycles calculated flow, this caused the sloshing in the pipes. In update 8 when the game paused to save the previous cycle data would effectively be lost and any wave action would fall flat. This helped some networks and hurt others so when 1.0 no longer paused the game to save some networks performed better and others did worse.
This is something that is so in the weeds I did not think anyone would benefit from it so I glazed over it.
Moving on to everything is a fluid buffer.
A lot of the fluid volumes are hidden to the player, which for most players is fine. Explaining the why is the hard part, the fluid mechanics is broken into two parts head lift which is used to limit what hight a liquid reach and fill percentage. These two processes interact with each other. head lift propagates through the network only through pipe/pipeline components which are full (100% or more full). The pipe network runs using floating point values with a soft limit of 140% full, this full range make the network act like a rubber hose but seams to be a necessary evil to prevent losing fluid in the network from sudden changes in flow. This is why everything has two values in the video a programed value and the overfilled value.
Each of the different components needs to have an internal volume, so the head lift can propagate through it.
The pipeline Junction programming was not adjusted when they added the ability to rotate it. That is why everything gets weird with it is rotated, all of that testing was just showing how I came to that.
The created several Reddit posts using this test stand in update 8, this was the first video where I included them.
I don’t have access to the program so some aspects I can’t 100% confirm.
I am a mechanical electrical control systems engineering and part of my IRL job has me reverse engineering control systems from the UI and IO. So take that into consideration.
I don’t think including the word salad above would have made the video clearer.