Overclock.net banner

1 - 20 of 24 Posts

·
Registered
Joined
·
4,261 Posts
Discussion Starter #1


I was hoping to emulate this configuration for my GPU loop but as I'm putting it together it's not clear to me how exactly the water would flow through both blocks to adequately cool them. Wouldn't the water just flow up to the top GPU and back down through the case without really going through the botttom GPU much if at all? Can someone explain to me what I'm obviously missing?
 

·
Banned
Joined
·
3,726 Posts
yes I can explain this parallel configuration.

http://www.overclock.net/forum/61-water-cooling/1573189-serial-vs-parallel-9-6lpm.html

http://www.overclock.net/forum/61-water-cooling/1615072-cpu-radiator-upgrade-water-cooled-rig.html

http://www.overclock.net/forum/62-peltiers-tec/1651377-full-system-tec-cooling.html

all I build today are parallel water loops for my systems.

now to answer that question, to start understand that what you are imaging is not exactly what happens.
so lets say that you have a water loop with these two video card water blocks in a parallel connection with each other. when the pump is turned on the water will start to flow but the flow rate through each water block will be the exact same flow rate because the resistance of the two water blocks are the same.

lets take it a little further, lets say that with one D5 pump the total flow rate of the water loop is 5LPM. that means that 2.5LPM is flowing through each water block.

now lets say that the total flow rate of the system with 2 D5 pumps is 8LPM. that means that now the flow rate through each water block is 4LPM.

it will always work this way, in fact I like a 3-way parallel connection for my builds, I will break it down for you.
on my serial vs parallel build my total flow rate was 9.6LPM. I know that the flow rate through the CPU was 3.6LPM because I have a flow meter in the loop right after the CPU. that means that the flow rate through each of the video card water blocks is 3.0LPM or the reaming flow rate divided by two.
 

·
Registered
Joined
·
4,261 Posts
Discussion Starter #3 (Edited)
yes I can explain this parallel configuration.

http://www.overclock.net/forum/61-water-cooling/1573189-serial-vs-parallel-9-6lpm.html

http://www.overclock.net/forum/61-water-cooling/1615072-cpu-radiator-upgrade-water-cooled-rig.html

http://www.overclock.net/forum/62-peltiers-tec/1651377-full-system-tec-cooling.html

all I build today are parallel water loops for my systems.

now to answer that question, to start understand that what you are imaging is not exactly what happens.
so lets say that you have a water loop with these two video card water blocks in a parallel connection with each other. when the pump is turned on the water will start to flow but the flow rate through each water block will be the exact same flow rate because the resistance of the two water blocks are the same.

lets take it a little further, lets say that with one D5 pump the total flow rate of the water loop is 5LPM. that means that 2.5LPM is flowing through each water block.

now lets say that the total flow rate of the system with 2 D5 pumps is 8LPM. that means that now the flow rate through each water block is 4LPM.

it will always work this way, in fact I like a 3-way parallel connection for my builds, I will break it down for you.
on my serial vs parallel build my total flow rate was 9.6LPM. I know that the flow rate through the CPU was 3.6LPM because I have a flow meter in the loop right after the CPU. that means that the flow rate through each of the video card water blocks is 3.0LPM or the reaming flow rate divided by two.

Thank you chiming in.

The only thing that I'm not following is how both GPU blocks in this instance have the same resistance. In a vacuum, yes they do. But in the case of the picture I posted, how is there just as much resistance in the lower block as there is in the upper block when the water is free to flow straight up the tubing to the upper block bypassing the lower block?

So I guess my real question is, what forces the water to actually circulate through the lower block?
 

·
Banned
Joined
·
3,726 Posts
Thank you chiming in.

The only thing that I'm not following is how both GPU blocks in this instance have the same resistance. In a vacuum, yes they do. But in the case of the picture I posted, how is there just as much resistance in the lower block as there is in the upper block when the water is free to flow straight up the tubing to the upper block bypassing the lower block?
well think of it this way if I fill a 5 gallon bucket with water and I have one big hole and one small hole does the water only flow through the big hole?

the answer is no, and I can show this if needed.

same thing if the holes are the same size the same amount of water will flow from each.
the same is true with the video card water blocks, the total flow rate will always divide itself between the two water blocks, or the holes that the water can flow through.
 

·
Registered
Joined
·
4,261 Posts
Discussion Starter #5 (Edited)
well think of it this way if I fill a 5 gallon bucket with water and I have one big hole and one small hole does the water only flow through the big hole?

the answer is no, and I can show this if needed.

same thing if the holes are the same size the same amount of water will flow from each.
the same is true with the video card water blocks, the total flow rate will always divide itself between the two water blocks, or the holes that the water can flow through.

Right but if you have a 5 gallon bucket and at the bottom of that bucket are two large holes and on the sides of that bucket are two small holes, isn't more water going to flow out of the larger holes as you poor water into the bucket? I would think so for two reasons:

1. Because the bottom holes are directly in line with where the water is coming from (above)
2. The holes are larger so there is less restriction for the water to flow out of them than the small holes to the side.

In the case of the lower block here, it seems like it's much easier for the water to flow straight up to the second block since the water is coming from below (in the case it's coming from the bottom left tube coming from the bottom of the case). And then coming back out of the top block now it's flowing down and would mostly bypass the bottom block to go back down through the bottom of the case (lower right tube).

You clearly know more about this than me so I will take your word on this, it's just not adding up in my head from a logical standpoint.
 

·
Banned
Joined
·
3,726 Posts
well ok, and I do understand, that you are right, but what you are seeing in your head, is only true for the first 10th of a sec. than the water flow levels out or equalizes among the two cards.

Right but if you have a 5 gallon bucket and at the bottom of that bucket are two large holes and on the sides of that bucket are two small holes, isn't more water going to flow out of the larger holes as you poor water into the bucket? I would think so for two reasons:

1. Because the bottom holes are directly in line with where the water is coming from (above)
2. The holes are larger so there is less restriction for the water to flow out of them than the small holes to the side.

In the case of the lower block here, it seems like it's much easier for the water to flow straight up to the second block since the water is coming from below (in the case it's coming from the bottom left tube coming from the bottom of the case). And then coming back out of the top block now it's flowing down and would mostly bypass the bottom blow to go back down through the bottom of the case (lower right tube).

You clearly know more about this than me so I will take your word on this, it's just not adding up in my head from a logical standpoint.
 

·
Registered
Joined
·
4,261 Posts
Discussion Starter #7
well ok, and I do understand, that you are right, but what you are seeing in your head, is only true for the first 10th of a sec. than the water flow levels out or equalizes among the two cards.

Right but if you have a 5 gallon bucket and at the bottom of that bucket are two large holes and on the sides of that bucket are two small holes, isn't more water going to flow out of the larger holes as you poor water into the bucket? I would think so for two reasons:

1. Because the bottom holes are directly in line with where the water is coming from (above)
2. The holes are larger so there is less restriction for the water to flow out of them than the small holes to the side.

In the case of the lower block here, it seems like it's much easier for the water to flow straight up to the second block since the water is coming from below (in the case it's coming from the bottom left tube coming from the bottom of the case). And then coming back out of the top block now it's flowing down and would mostly bypass the bottom blow to go back down through the bottom of the case (lower right tube).

You clearly know more about this than me so I will take your word on this, it's just not adding up in my head from a logical standpoint.
I see. So once the tubes are fully populated with water the flow will even out is what you're saying. I see how that would work. Thank you for talking me through this.
 

·
Banned
Joined
·
3,726 Posts
your welcome.
 

·
Registered
Joined
·
1,558 Posts
I love being on OCN. I would rep Tool if I could. I was also a bit confused. Props man :).
 

·
Registered
Joined
·
1,105 Posts
I see. So once the tubes are fully populated with water the flow will even out is what you're saying. I see how that would work. Thank you for talking me through this.
Just to elaborate on this a little more, hopefully it will click more.

Water is not compressible, (not much if any) So you have to factor in pressure from the pump/s and that the system is sealed (hopefully)

So in parallel, as said above, fluid will be halved through the two blocks. In series, the flow rate will remain the same.

Dual pump configs, flow rate will be half of whatever the system is pushing in parallel, in series head pressure will double. (hopefully that makes sense) But fluid is having pressure applied and wants to equalize, so it will still flow through the lower block just the same as the upper block, but at half the flow rate of the rest of the system.

The bucket examples above, (if it was sealed) the water would flow out of all four holes except the velocity will be faster from the smaller holes, but volume will be greater through larger holes.
 

·
Premium Member
Joined
·
2,812 Posts
The bottom GPU does get more flow. It's a point of technicality that isn't significant in the setup shown, but would matter greatly in large scale systems.

At zero gravity, they get the same amount of flow. On Earth, there will be greater pressure at lower points. Gravity impacts the pressure, and flow takes the path of least resistance. If the standing pipe between the two GPU's were 100 feet long vertical, that'd create a pressure difference of roughly 43 PSI -- the top GPU would get almost no flow at all because 43 psi of resistance would need to be overcome in order to reach that top GPU; the path of least resistance would be in through the inlet and then out of the discharge of the bottom GPU.

A difference of 3 inches in a real system though? Wouldn't matter. I just want you all to know.. I'm smarter. ;)
 

·
Registered
Joined
·
4,261 Posts
Discussion Starter #12
Just to elaborate on this a little more, hopefully it will click more.

Water is not compressible, (not much if any) So you have to factor in pressure from the pump/s and that the system is sealed (hopefully)

So in parallel, as said above, fluid will be halved through the two blocks. In series, the flow rate will remain the same.

Dual pump configs, flow rate will be half of whatever the system is pushing in parallel, in series head pressure will double. (hopefully that makes sense) But fluid is having pressure applied and wants to equalize, so it will still flow through the lower block just the same as the upper block, but at half the flow rate of the rest of the system.

The bucket examples above, (if it was sealed) the water would flow out of all four holes except the velocity will be faster from the smaller holes, but volume will be greater through larger holes.

What are the consequences of of the GPU blocks having half the flow rate? My GPU loop will have a 840mm (560 + 280 SR2's) of rad space so I imagine I'll still be fine on the cooling front.


The bottom GPU does get more flow. It's a point of technicality that isn't significant in the setup shown, but would matter greatly in large scale systems.

At zero gravity, they get the same amount of flow. On Earth, there will be greater pressure at lower points. Gravity impacts the pressure, and flow takes the path of least resistance. If the standing pipe between the two GPU's were 100 feet long vertical, that'd create a pressure difference of roughly 43 PSI -- the top GPU would get almost no flow at all because 43 psi of resistance would need to be overcome in order to reach that top GPU; the path of least resistance would be in through the inlet and then out of the discharge of the bottom GPU.

A difference of 3 inches in a real system though? Wouldn't matter. I just want you all to know.. I'm smarter. ;)

Thanks for putting it in laymens terms for the lesser minds amongst us :thumb:.
 

·
Banned
Joined
·
3,726 Posts
What are the consequences of of the GPU blocks having half the flow rate? My GPU loop will have a 840mm (560 + 280 SR2's) of rad space so I imagine I'll still be fine on the cooling front.
.
I would not look at it as half in that regard as the flow rate will be twice the flow rate of one.

let me give you a example of what I mean.

lets say that you build a simple water loop with one pump, one reservoir, and the GPU water block in the water loop. now lets say that the total flow rate through that loop is 4LPM.

now lets say that I build a simple water loop with one pump, one reservoir, and two GPU water blocks in parallel with each other, the flow rate on this water loop will be from 7LPM to 8LPM.

now comes the part where you will actually have to test what the flow rate through one of your radiators is. my radiators on a simple water loop with one pump, one reservoir, and one radiator is only 5.9LPM. so I had to parallel my radiators as well. so that I would be able to achieve a high enough flow rate on my water loop, to provide my components with enough flow for each of them to stay nice and cool.
 

·
Registered
Joined
·
4,261 Posts
Discussion Starter #14
What are the consequences of of the GPU blocks having half the flow rate? My GPU loop will have a 840mm (560 + 280 SR2's) of rad space so I imagine I'll still be fine on the cooling front.
.
I would not look at it as half in that regard as the flow rate will be twice the flow rate of one.

let me give you a example of what I mean.

lets say that you build a simple water loop with one pump, one reservoir, and the GPU water block in the water loop. now lets say that the total flow rate through that loop is 4LPM.

now lets say that I build a simple water loop with one pump, one reservoir, and two GPU water blocks in parallel with each other, the flow rate on this water loop will be from 7LPM to 8LPM.

now comes the part where you will actually have to test what the flow rate through one of your radiators is. my radiators on a simple water loop with one pump, one reservoir, and one radiator is only 5.9LPM. so I had to parallel my radiators as well. so that I would be able to achieve a high enough flow rate on my water loop, to provide my components with enough flow for each of them to stay nice and cool.

I don't intend to run my rads in parallel but I am using a D5 strong in both my loops. I wasn't intending on running them at full power but I can if I need more flow.
 

·
Registered
Joined
·
1,165 Posts
The bottom GPU does get more flow. It's a point of technicality that isn't significant in the setup shown, but would matter greatly in large scale systems.

At zero gravity, they get the same amount of flow. On Earth, there will be greater pressure at lower points. Gravity impacts the pressure, and flow takes the path of least resistance. If the standing pipe between the two GPU's were 100 feet long vertical, that'd create a pressure difference of roughly 43 PSI -- the top GPU would get almost no flow at all because 43 psi of resistance would need to be overcome in order to reach that top GPU; the path of least resistance would be in through the inlet and then out of the discharge of the bottom GPU.

A difference of 3 inches in a real system though? Wouldn't matter. I just want you all to know.. I'm smarter. ;)

Thats true of a point to point system but in a loop like pc cooling systems gravity essentially plays no role. The vertical head resistance is counteracted by the positive pressure on the pumps inlet from the other side of the loop.




I was hoping to emulate this configuration for my GPU loop but as I'm putting it together it's not clear to me how exactly the water would flow through both blocks to adequately cool them. Wouldn't the water just flow up to the top GPU and back down through the case without really going through the botttom GPU much if at all? Can someone explain to me what I'm obviously missing?


The pictured connections work because pumping does not rely on inertia of the fluid to determine its path. Rather then thinking of one path as easier to get too or something like that you need to always think of pressure differentials because that is what causes flow, nothing more.
The vertical tube on the left is recieving water from the pump so its under positive pressure. The water wants to escape and has two paths to get over to the right hand tube. These paths are identical water blocks so their flow resistance is equal, which causes their flow rates to be equal.
If one path was twice the restriction then the lower resistance path would get about 50% more flow than the other. So it takes very large differences in flow paths for there to be significant differences in flow rate.


Someone really does make every single post about themselves don't they
 

·
Banned
Joined
·
3,726 Posts
I don't intend to run my rads in parallel but I am using a D5 strong in both my loops. I wasn't intending on running them at full power but I can if I need more flow.
that is a great way of looking at the D5 strong pumps, I used two D5 strong pumps in my build, but I could had gotten away with just using one. the reason I used two, is because I wanted to get my flow rate up to 12LPM, not because I had to, but because I wanted to. I used to build my cars up, but today I am to old for that, so now I build my water cooling systems up, I still have that want to always go faster.
 

·
Premium Member
Joined
·
2,812 Posts
Thats true of a point to point system but in a loop like pc cooling systems gravity essentially plays no role. The vertical head resistance is counteracted by the positive pressure on the pumps inlet from the other side of the loop.
I clarified that in the original post. My post was blatantly overly technical to a point of absurdity, given the setup -- and I said so. Then, at the end of the post, I even sarcastically mocked my own motives for chiming in at all.

Ashcroft said:
Someone really does make every single post about themselves don't they
So if this is directed at me, relax. I made it abundantly clear that what I was saying 1.) didn't actually apply, and 2.) displayed pretentiousness.

I did the work for you. Whacking me over the head, at this point, is just in bad taste.
 

·
Registered
Joined
·
911 Posts
So if this is directed at me, relax. I made it abundantly clear that what I was saying 1.) didn't actually apply, and 2.) displayed pretentiousness.

I did the work for you. Whacking me over the head, at this point, is just in bad taste.
Knowing Ashcroft it was not directed at you. There is a lot of history behind his comment. I do find in interesting that you would think it was.
 
1 - 20 of 24 Posts
Top