Archived

This topic is now archived and is closed to further replies.

Please be aware that the content of this thread may be outdated and no longer applicable.

Chthonicone

Why Liquid Loops Stop When Full

Recommended Posts

Chthonicone    89

I have discovered why liquid loops stop when you fill them with too much liquid. I'm not sure if the exact reason has been found by anyone else, but I'm pretty sure there are at least a few people doing cooling builds looking for this reason.

First, how does a cooling loop work? For the loop to work, the liquids need to be able to move from a outlet to an inlet somewhere on the loop. If there is no inlet to flow to on a path, it will not take that path. This is why most cooling loops have a liquid bridge on them, though anything with an outlet and inlet that lets the liquid through can function for this. E.g. valves, shutoffs, some refining machines.

Now lets look at how these devices work. If their outlet is not blocked, any liquid that begins entering the pipe on the inlet is immediately transported to the pipe on the outlet if it is not blocked. If the outlet pipe is connected to an inlet pipe somewhere, it will begin moving the liquid in the same second. In this case, the liquid never actually was in the inlet pipe. Hover your mouse over it as the liquids flow, and you'll see that fact. That pipe always remains empty.

If the outlet is blocked though, the liquid that enters the inlet pipe cannot be transported, so it stays. As such, an extra packet of liquid can enter the system somewhere to fill the gap created by allowing us to hide this packet of liquid.

This is where the problem begins. If there are no empty pipes in the loop, liquid will not move. At least one packet of liquid needs to see that there is room in front of it to move up. If all such gaps caused by the inlet pipes are filled, the loop will never flow again. In fact, if some are filled but not others, you may see packets start and stop in phases around the loop. This indicates that at least one inlet pipe cannot get rid of it's packet before all the other inlets do. Now the gaps are traded around the loop in a round-robin fashion. This is not optimal.

Some people I've read have claimed that adding another gap into the loop can fix this, but this is only true if  you have stopped adding more liquid and it is not optimal unless you were off by one packet.

In fact, the only way I can see right now to fix this problem is to first remove a bridge or valve somewhere, liquids are never stored in these and will not cause your system to leak. Then you count the total number of bridges and valves in your loop, and step ahead that many segments of pipe and have a plumber empty from there. This is ahead of the piece you've made, not behind it. This way you remove the exact amount of packets you need to. Finally, reinstall your bridge and it should start running optimally.

There is a way to safe fill this up to capacity though. This involves using a liquid bridge to create a filler valve, as in the bridge exits directly into the cooling loop, and then fill that way. So long as the loop can flow naturally during the filling process, you wait till it's full, then remove the bridge.

I hope this helps those who were confused about cooling loops and why they stop on you.

Share this post


Link to post
Share on other sites
beowulf2010    537

The only time I've ever had loops fail on me is due to operator errors such as pipe T's or multiple white connections in a single run of piping...

Share this post


Link to post
Share on other sites
fragtzack    9

TonyAdvanced has a great double bridge design buffer space device that is pretty fool proof to prevent a loop from backing up. youtube/google it.

Share this post


Link to post
Share on other sites
Chthonicone    89
3 hours ago, Steve Raptor said:

I'm using aqua tuner cooling loops consistently and I never had this problem, I use liquid reservoirs as buffers and thats it.

 

3 hours ago, beowulf2010 said:

The only time I've ever had loops fail on me is due to operator errors such as pipe T's or multiple white connections in a single run of piping...

Then this is here to alert you that there are pitfalls to the problem, and how to avoid them if you do encounter them.

2 hours ago, fragtzack said:

TonyAdvanced has a great double bridge design buffer space device that is pretty fool proof to prevent a loop from backing up. youtube/google it.

Show me the setup and I'll tell you how or if it can fail. Likely there is something he does to prevent you from ever overfilling the loop.

Share this post


Link to post
Share on other sites
Majestix    249

You are making quite a big deal out of this. Just have a buffer tank in your cooling loop, like Steve Raptor said, and don't fill it up completely. Of course, you still can break it by completely filling up the loop and the tank, but if you are not actually trying to break it, it will just work.

I just disconnect the coolant supply manually when I am happy with my loop, but if you really want to automate it and leave the supply pipe in place, then put a liquid sensor in the last segment of the cooling pipe just before it loops back to the supply point (I would connect the supply to the in port of the buffer tank, and then place the sensor on the last segment of the cooling loop before the coolant comes back into the tank), have a liquid shutoff in the supply line, and deactivate the supply when the sensor detects any liquid. You can fill a bit of the buffer tank by either inserting a delay buffer gate in the automation or having the shutoff further down the supply pipe. Having some extra coolant in the buffer tank helps stabilise the temperature of the coolant.

Share this post


Link to post
Share on other sites
BLACKBERREST3    234
12 hours ago, fragtzack said:

TonyAdvanced has a great double bridge design buffer space device that is pretty fool proof to prevent a loop from backing up. youtube/google it.

You might want to check your facts, this story was told in another topic. It mentions a lot of the mechanics behind the AT like a packet blockage and the first packet slipping. The only downside to the double bridge is the stagnant single packet that is left in the left input of the bridge. I also mention certain use cases where a double bridge can be used instead of the usual buffer reservoir.

 

Spoiler

5d88a33717508_AquatunerBypass.png.d0b359c8931b9fed9f6bf7fadb2d8708.png5d88a337cc6a2_AquatunerBypassLaidOut.png.59e739c22790daccecf59ea721243d26.png

I'm just gonna throw in ONI University if you haven't checked it out yet. I'm still updating it. 

 

 

Share this post


Link to post
Share on other sites
Chthonicone    89
4 hours ago, BLACKBERREST3 said:

You might want to check your facts, this story was told in another topic. It mentions a lot of the mechanics behind the AT like a packet blockage and the first packet slipping. The only downside to the double bridge is the stagnant single packet that is left in the left input of the bridge. I also mention certain use cases where a double bridge can be used instead of the usual buffer reservoir.

 

  Reveal hidden contents

5d88a33717508_AquatunerBypass.png.d0b359c8931b9fed9f6bf7fadb2d8708.png5d88a337cc6a2_AquatunerBypassLaidOut.png.59e739c22790daccecf59ea721243d26.png

I'm just gonna throw in ONI University if you haven't checked it out yet. I'm still updating it. 

 

 

Wait? Tonyroid = Tony Advanced? I didn't realize they were the same person. In his videos even he suspects his solution might actually fail at some point, and it was him I talked to about possibly using a whole reservoir instead of a single pipe buffer. I was honestly hoping he could find this and see it because this explains the problem he had been having.

Share this post


Link to post
Share on other sites
BLACKBERREST3    234

I kinda figured it was him because of his profile pic. mathmanican didn't correct me on it when we worked on oni university. he also says that francis john is john francis, I'm taking his word for it. What's funny is that tony had a video up the first time he posted the topic. I'd rather he left it unlisted and not delete it unless you are referring to precisely that video, but I think that topic pretty much covered the basics and then some. As a side note, It would be pretty easy to impersonate a youtuber who doesn't visit the forums that often...cough @Brothgar cough cough.

Edit. ok. just watched his vid, I know what your talking about now.

Spoiler

 

ZanthraSW has the solution to this I thought. Still haven't tested this, but it's supposed to get rid of the stagnant packet by alternating the reservoir pipe so the packets keep moving. Also, I'm pretty sure the valve doesn't have to be in his build if he's using the double bridge bypass.

-Zan's Build

Spoiler

2019-09-10.thumb.png.6ee280cb35fdcaa06ed435eaec60b589.png.5df3d444e2fe2d6c460d5cc0dc47ea0d.png

 

Share this post


Link to post
Share on other sites
Chthonicone    89
24 minutes ago, BLACKBERREST3 said:

I kinda figured it was him because of his profile pic. mathmanican didn't correct me on it when we worked on oni university. he also says that francis john is john francis, I'm taking his word for it. What's funny is that tony had a video up the first time he posted the topic. I'd rather he left it unlisted and not delete it unless you are referring to precisely that video, but I think that topic pretty much covered the basics and then some. As a side note, It would be pretty easy to impersonate a youtuber who doesn't visit the forums that often...cough @Brothgar cough cough.

Edit. ok. just watched his vid, I know what your talking about now.

  Hide contents

 

ZanthraSW has the solution to this I thought. Still haven't tested this, but it's supposed to get rid of the stagnant packet by alternating the reservoir pipe so the packets keep moving. Also, I'm pretty sure the valve doesn't have to be in his build if he's using the double bridge bypass.

-Zan's Build

  Hide contents

2019-09-10.thumb.png.6ee280cb35fdcaa06ed435eaec60b589.png.5df3d444e2fe2d6c460d5cc0dc47ea0d.png

 

That is the video I talked to him about, including possibly using reservoirs to do the job, but as I said, reservoirs, like his buffer, just puts the problem off. The trick is to get exactly enough packets total into the system.

 

Not sure what exactly the second image accomplishes though that a simpler buffer wouldn't do already. I mean it just adds 3 more packets of buffer is all, to different places. If they all fill up, it'll stop again.

Share this post


Link to post
Share on other sites
BLACKBERREST3    234

Nope, just stress tested it, it does not fill up. toggle it on and off as much as you'd like. No stagnant packets or pipe blockages. I hope Zan likes the name cause I dub it the Quadruple Bridge Bypass.

5d89471d9742c_AquatunerBypassQuadrupleBridge.thumb.png.f0f5334f76b11650b6376d2a147c206a.png

 

Wait it can be improved!

Spoiler

2019-09-10.thumb.png.6ee280cb35fdcaa06ed435eaec60b589.thumb.png.c1a6e69a4f2fe492870c7e6de7990d82.png

:p

 

Maybe I'm not getting what you're trying to accomplish, but this build lets you hook up your liquid piping and let it stay connected. You could just as easily remove the liquid input and have a closed system, but this just gives it flexibility say if you wanted to filter liquid out by temperature, germs, or type.

Share this post


Link to post
Share on other sites
fragtzack    9

The Tony double bridge device I was referring to is not the actual bypass, the double bridge thing being shown here by Blackberry is the actual bypass. Not quite the same thing.

The double bridge system Tony shows does work for me along with his bypass method of using a liquid valve. Not a single blockage in my current game and just broke 1000 cycles.

In a closed loop (which I found is usually best anyways), most of this is pointless talk because all you need to do insure 100% uptime is put a single empty packet per aqua tuner in the loop.

 

Double bride in the loop to add a single empty packet space, Connect the 2 inputs to each other and the 2 exits to each other.  This is not the bypass, rather inline as  part of the loop.

 

image.png.95ae9180dc13a06139124fd061de4b77.png

 

Like to add, that Tony did go back and post a correction video based on discussion here. His double bridge system for a buffer space didn't change, but his bypass did..just use a liquid valve so the bypass pipe is the same length as going through the AT.

 

Share this post


Link to post
Share on other sites
Chthonicone    89
3 hours ago, BLACKBERREST3 said:

Nope, just stress tested it, it does not fill up. toggle it on and off as much as you'd like. No stagnant packets or pipe blockages. I hope Zan likes the name cause I dub it the Quadruple Bridge Bypass.

5d89471d9742c_AquatunerBypassQuadrupleBridge.thumb.png.f0f5334f76b11650b6376d2a147c206a.png

 

Wait it can be improved!

  Reveal hidden contents

2019-09-10.thumb.png.6ee280cb35fdcaa06ed435eaec60b589.thumb.png.c1a6e69a4f2fe492870c7e6de7990d82.png

:p

 

Maybe I'm not getting what you're trying to accomplish, but this build lets you hook up your liquid piping and let it stay connected. You could just as easily remove the liquid input and have a closed system, but this just gives it flexibility say if you wanted to filter liquid out by temperature, germs, or type.

 

2 hours ago, fragtzack said:

The Tony double bridge device I was referring to is not the actual bypass, the double bridge thing being shown here by Blackberry is the actual bypass. Not quite the same thing.

The double bridge system Tony shows does work for me along with his bypass method of using a liquid valve. Not a single blockage in my current game and just broke 1000 cycles.

In a closed loop (which I found is usually best anyways), most of this is pointless talk because all you need to do insure 100% uptime is put a single empty packet per aqua tuner in the loop.

 

Double bride in the loop to add a single empty packet space, Connect the 2 inputs to each other and the 2 exits to each other.  This is not the bypass, rather inline as  part of the loop.

 

image.png.95ae9180dc13a06139124fd061de4b77.png

 

Like to add, that Tony did go back and post a correction video based on discussion here. His double bridge system for a buffer space didn't change, but his bypass did..just use a liquid valve so the bypass pipe is the same length as going through the AT.

 

I have had similar systems fail during reload, are you certain there are no reload shenanigans that can happen?

Share this post


Link to post
Share on other sites
BLACKBERREST3    234
1 hour ago, Chthonicone said:

The Tony double bridge device I was referring to is not the actual bypass, the double bridge thing being shown here by Blackberry is the actual bypass. Not quite the same thing.

My apologies, I almost overlooked it. I combined The original from saturnus with the the inline double bridge tony used. Now we have the Triple Bridge Bypass. The added benefit of having 1 less bridge. I can't be certain it survives save/reload, but I'll let you know if it fails at any point. This does the exact same thing as the quadruple bridge.

Spoiler

5d8982a166f74_ATTripleBridgeBypass.thumb.png.5af57f61fd3e07282e4d2a6d06c452d0.png

Nevermind, took a little bit to break it. It still suffers from stagnant packet on 100% uptime if you flip it on and off just right.

Spoiler

5d8987555b7b0_StagnantPackets.thumb.png.2ff2791f3385de18e09ef083be058911.png

Share this post


Link to post
Share on other sites
TLW    56

Adding more bridges just disguises the problem.

The aquatuner has state that most people don't think about - it can either be filled with liquid, or it can be empty. And depending on exactly when you turn on/off the aquatuner, you can end up in either state.

The main issue I've seen is "overfilling". In quotes because it's not quite traditional overfilling.

You're safe if you're using something that has at least one (two?) free space(s) of "buffer", and you ensure that no more liquid enters the loop while there's not enough space in the buffer.

What this means is roughly as follows:

  1. The traditional single-bridge bypass isn't robust. If the aquatuner turns off (on?) at the wrong time, you can end up with it being deadlocked due to its internal buffer.
  2. The "standard" double-bridge bypass (e.g. image.png.3dfda08ea3b2fb719cad8305fb72e0 ) is only robust if you fill it entirely with the aquatuner having never been run, remove the input bridge, and then (and only then) enable the aquatuner.
  3. A double-bridge bypass with the liquid input inputting into the bypass buffer (e.g. image.png.ff18e94820ac811e16fe38e13b0a64, but with the input bridge instead going to the empty space at the bottom left) is robust. Well, should be.

That being said, this is mainly theoretical. I personally have given up and have started including a reservoir into my builds. It have become very apparent that "elegant" designs can't be relied on in ONI. If not today then tomorrow or the day after. Saddening...

Share this post


Link to post
Share on other sites
BLACKBERREST3    234

From my testing (flipping a switch for ten minutes) , the quadruple bridge by zanthrasw does not overfill regardless of what state the AT is in. Neither does the valve with a double bridge buffer that tony used. @TLW, do these builds overfill or have stagnant packets at any time even during a save/reload for you? I wouldn't mind help with testing, it can be fickle.

Share this post


Link to post
Share on other sites
Lancar    85
On 9/22/2019 at 11:50 PM, Steve Raptor said:

I'm using aqua tuner cooling loops consistently and I never had this problem, I use liquid reservoirs as buffers and thats it.

Another upside to using a tank like this is that it enables you to drain the system at any time you need to re-wire or otherwise service it, without any leakage (done by Disabling the reservoir. It'll then accept inputs but not send anything to the output).

Share this post


Link to post
Share on other sites
Tonyroid    254
On 9/22/2019 at 2:44 PM, Chthonicone said:

I have discovered why liquid loops stop when you fill them with too much liquid. I'm not sure if the exact reason has been found by anyone else, but I'm pretty sure there are at least a few people doing cooling builds looking for this reason.

First, how does a cooling loop work? For the loop to work, the liquids need to be able to move from a outlet to an inlet somewhere on the loop. If there is no inlet to flow to on a path, it will not take that path. This is why most cooling loops have a liquid bridge on them, though anything with an outlet and inlet that lets the liquid through can function for this. E.g. valves, shutoffs, some refining machines.

Now lets look at how these devices work. If their outlet is not blocked, any liquid that begins entering the pipe on the inlet is immediately transported to the pipe on the outlet if it is not blocked. If the outlet pipe is connected to an inlet pipe somewhere, it will begin moving the liquid in the same second. In this case, the liquid never actually was in the inlet pipe. Hover your mouse over it as the liquids flow, and you'll see that fact. That pipe always remains empty.

If the outlet is blocked though, the liquid that enters the inlet pipe cannot be transported, so it stays. As such, an extra packet of liquid can enter the system somewhere to fill the gap created by allowing us to hide this packet of liquid.

This is where the problem begins. If there are no empty pipes in the loop, liquid will not move. At least one packet of liquid needs to see that there is room in front of it to move up. If all such gaps caused by the inlet pipes are filled, the loop will never flow again. In fact, if some are filled but not others, you may see packets start and stop in phases around the loop. This indicates that at least one inlet pipe cannot get rid of it's packet before all the other inlets do. Now the gaps are traded around the loop in a round-robin fashion. This is not optimal.

Some people I've read have claimed that adding another gap into the loop can fix this, but this is only true if  you have stopped adding more liquid and it is not optimal unless you were off by one packet.

In fact, the only way I can see right now to fix this problem is to first remove a bridge or valve somewhere, liquids are never stored in these and will not cause your system to leak. Then you count the total number of bridges and valves in your loop, and step ahead that many segments of pipe and have a plumber empty from there. This is ahead of the piece you've made, not behind it. This way you remove the exact amount of packets you need to. Finally, reinstall your bridge and it should start running optimally.

There is a way to safe fill this up to capacity though. This involves using a liquid bridge to create a filler valve, as in the bridge exits directly into the cooling loop, and then fill that way. So long as the loop can flow naturally during the filling process, you wait till it's full, then remove the bridge.

I hope this helps those who were confused about cooling loops and why they stop on you.

I saw this thread when it went up but I didn't read it until now. I would definitely say your analysis is correct.

The thing I find tricky about preventing deadlock in loops is that loops come in many varieties and have lots of different requirements.Simply filling them to capacity and then disconnecting them from an input isn't always an option. Sometimes having an exit and an entrance is intrinsic to the purpose of the loop.

I find that making sure the point where new liquid is inserted into a loop comes immediately after a buffer space is a good way to make sure that the loop doesn't get over filled. But that only works if the buffer is a configuration that is compatible with the loop. That seems to be the hard part.

I have been able to prove that it's possible to make loops that require no less than two reservoirs arranged in parallel to ensure that no deadlocks can possibly occur. At this time, I think that might be adequate for ANY loop, but I consider it too extensive to seriously consider for loops that don't need that much. So, I don't think there's a good universal solution to the problem. Instead, every loop needs to be evaluated independently.

I made a loop today that tricks a reservoir/buffer into outputting a partial packet, even if the buffer is full. I didn't even know that was possible. This loop literally requires a reservoir AND another small buffer to prevent deadlock.

Share this post


Link to post
Share on other sites
BLACKBERREST3    234
45 minutes ago, Tonyroid said:

every loop needs to be evaluated independently.

I can attest to this. It all depends on what your goal is and what type of loop you are willing to settle for be it simple or complex. 

 

46 minutes ago, Tonyroid said:

I made a loop today that tricks a reservoir/buffer into outputting a partial packet, even if the buffer is full.

cool, we'd love to see it.

Share this post


Link to post
Share on other sites
Yunru    1,139
4 hours ago, Tonyroid said:

Instead, every loop needs to be evaluated independently.

Couldn't you just tie a shut-off to a sensor so it only sends 1 packet every 2 ticks? 

... 

Thinking about it, is there a reason a valve set to 50% doesn't work? 

Share this post


Link to post
Share on other sites
Steve Raptor    210

Here is my cooling loop design, I have been using this extensively since the re-work of the steam turbine and it never had any problems whatsoever.

Theory of operation:

  • Both thermo sensors are set to the same value and "Above", this determines the desired temperature of the coolant.
  • Coolant exits the left reservoir and is checked for temperature, if its below the set value on the thermo sensor the liquid shut off will not activate and the coolant is sent into the cooling pipes.
  • If the coolant temperature is higher than the set value then the liquid shutoff valve will activate and coolant will be sent into the aqua tuner cooling loop through the right reservoir.
  • The coolant coming out of the aqua tuner gets priority to enter the cooling pipes over the one coming out of the liquid reservoir, the output of the aqua tuner loop is never stagnant and is either empty or flowing.
  • The coolant ends up in the left liquid reservoir after it passes through the cooling pipes.
  • The process repeats.

Another neat feature is that you can easily replace the coolant by deactivating the liquid reservoirs, this will drain the coolant from the system, allowing you to replace it with a different one.

In the Working example, I had polluted water running for a long time that got recently replaced by super coolant.

Base Design:

Spoiler

514638E4E5EAF0DE81A6367787B7C4A294976F5D

 

Working Example:

Spoiler

429D3F60AC21CD55BA80D1CD2E7EDEA40E906DF1

 

Maybe this is the solution your all looking for?

 

 

 

Share this post


Link to post
Share on other sites
TaeganRiles    2

Here's the solution I've come up with.  I'm using two Liquid Reservoirs anywhere I want to put a liquid loop then piping them in a way I believe ensures consistent flow through the loop and prevents the system from being overfilled.

I'm calling this build a Liquid Loop Regulator.  I've tried to build it so it can maintain a local loop, send packets elswhere as required, and accept packets back into the the local loop as needed.

I've marked the first attached image to hopefully explain how it's setup.  There is a Primary reservoir where thermal averaging happens and an Overflow reservoir.

A bridge between the outputs of the reservoirs combines their outputs into one continuous flow that goes to a radiator or Aquatuner loop.  Connecting the two reservoirs together into one bridge creates a bottleneck that is necessary for the system to function as intended.

The radiator/Aquatuner loop comes back to the Primary reservoir or overflows as needed into the Overflow reservoir. 

The Output line is for sending packets out of the local loop for thermal exchange elsewhere and the Return is where liquid can be bridged back into the local loop.

Hopefully this helps.

Liquid Loop Regulator with notes.png

Liquid Loop Regulator - Example 1 - Empty.PNG

Liquid Loop Regulator - Example 2 - Filling Up.PNG

Liquid Loop Regulator - Example 3 - Thermal tank full.PNG

Share this post


Link to post
Share on other sites
TaeganRiles    2

Depends on what your goal is I suppose, but why dump liquid out of a cooling loop and spend extra power to pump it back in when you can use reservoirs to do effectively the same thing without consuming additional power?

Share this post


Link to post
Share on other sites
TaeganRiles    2

So after letting my test lab run for a while I discovered how my setup handles a backup on the output, it empties the local loop.  This is a good thing if continuing to run the local loop during a crisis might result in pipe damage.  I also realized that I unintentionally build a failsafe in with how I have the liquid shutoffs positioned; if the power ever fails the contents of the system will dump out of the loops and just cycle between the overflow reservoirs.  This should prevent any damage to pipes from stagnant liquid freezing or boiling.  One less thing to worry about when taking care of a failure elsewhere.

Liquid Loop Regulator - At Capacity.PNG

Share this post


Link to post
Share on other sites