Non Traditional Flow through Radiator

[Zarathustra[H]]

Hey all,

I am plotting the flow for a new water loop, and I could use some [H] hive mind input.


I have an Alphacool XT45 420mm radiator, I am looking to repurpose into the new loop.

Normally one would use the following ports as in and out:

Is there any reason not to do something like this?

It would make my routing a lot simpler if this isn't a problem, but if it does result in issues, I may have to figure something else out.

Appreciate any input.

 

[thesmokingman]

That's a vent/fill port, not for flow. Flow path doesn't work like that. What you are wanting is a crossflow rad. Regular flow path rad, that vent port would render half the rad useless or another way of looking at it cut the rad in half. Water will travel thru the path of least resistance. That said I've never tried this so it might be fun for the sake of science.

 

[Zarathustra[H]]

That's a vent/fill port, not for flow. Flow path doesn't work like that. What you are wanting is a crossflow rad. Regular flow path rad, that vent port would render half the rad useless or another way of looking at it cut the rad in half. Water will travel thru the path of least resistance. That said I've never tried this so it might be fun for the sake of science.

Yeah, I'm aware the intent is for that port to be a vent port. I'm curious if anyone has tried flowing through it though.

I understand that would result in slower flow through the radiator, but I'm not convinced that is a terrible thing.

 

[thesmokingman]

Yeah, I'm aware the entent is for that port to be a vent port. I'm curious if anyone has tried flowing through it though.

I understand that would result in slower flow through the radiator, but I'm not convinced that is a terrible thing.

Nah, yer not quite getting it. GIS a 2 pass radiator. You're cutting the rad in half essentially. Flow will then not go thru the traditional cold side, it would just flow thru the shortest least resistant path, all thru the hot side. Btw, I'm guessing the nexxos is a tradition 2 pass rad. If its a 2 pass like HWL... can of worms.

 

[VanGoghComplex]

The water will not all flow through the path of least resistance. It will be a ratio defined by inverses. Think of the equation for calculating the resistance of parallel resistors.

The only "greater resistance" one half will have over the other is the fact that the vent port is located off the center axis of the rad. I'm betting that difference will effectively be within the margin of error if you were to actually test it.

Furthermore, the flow resistance offered by the radiator will be half of what it is in the normal U configuration. Instead of forcing water through 8 channels for 720mm with a sharp 180° turn in the middle, you'd be forcing water through 16 channels for only 360mm and the flow direction would only ever change by 90°.

As long as you split your flow at the terminal end like your diagram shows so that your water is going into both sides of the "U", you'll effectively have an x-flow rad.

 

[BrotherMichigan]

The water will not all flow through the path of least resistance. It will be a ratio defined by inverses. Think of the equation for calculating the resistance of parallel resistors.

Quoted for emphasis. I don't know where this belief that flow only follows the path of least resistance came from, but I see it everywhere.

I understand that would result in slower flow through the radiator, but I'm not convinced that is a terrible thing.

All other things being equal, more flow = better performance. I and many others have explained why this is enough times in various and sundry places that I won't bother taking the time to restate it here. If you're curious, a bit of searching will lead you to explanations as to why slower flow through blocks and radiators is in fact NOT beneficial to cooling performance.

 

[Zarathustra[H]]

The water will not all flow through the path of least resistance. It will be a ratio defined by inverses. Think of the equation for calculating the resistance of parallel resistors.

The only "greater resistance" one half will have over the other is the fact that the vent port is located off the center axis of the rad. I'm betting that difference will effectively be within the margin of error if you were to actually test it.

That's a vent/fill port, not for flow. Flow path doesn't work like that. What you are wanting is a crossflow rad. Regular flow path rad, that vent port would render half the rad useless or another way of looking at it cut the rad in half. Water will travel thru the path of least resistance. That said I've never tried this so it might be fun for the sake of science.

Furthermore, the flow resistance offered by the radiator will be half of what it is in the normal U configuration. Instead of forcing water through 8 channels for 720mm with a sharp 180° turn in the middle, you'd be forcing water through 16 channels for only 360mm and the flow direction would only ever change by 90°.

As long as you split your flow at the terminal end like your diagram shows so that your water is going into both sides of the "U", you'll effectively have an x-flow rad.

That's kind of what I figured.

I suspected he offset of the port might result in different flows on different sides of the radiator, but I was uncertain to what extent.

Another problem I anticipated was since the port is not designed for flow, it may not be optimized and as such might be more restrictive. No idea though.

I can probably just check how uneven the flow would be with a funnel and two buckets. Pour water into the side with the one port, and see if there is a significantn difference in the amount of water that winds up in each bucket.

Actually, I'd need something more steady state than that. Maybe hook it up to a faucet.

 

[Zarathustra[H]]

Quoted for emphasis. I don't know where this belief that flow only follows the path of least resistance came from, but I see it everywhere.



All other things being equal, more flow = better performance. I and many others have explained why this is enough times in various and sundry places that I won't bother taking the time to restate it here. If you're curious, a bit of searching will lead you to explanations as to why slower flow through blocks and radiators is in fact NOT beneficial to cooling performance.

I am well aware that you want a high flow to even out temperature differences across a block/radiator.

In this case though the comparison winds up being between one U-shaped higher flow radiator used as designed, vs essentially turning it into two shorter and smaller radiator segments with lower flow locally only over that radiator.

I suspect the difference between the two would be negligible.

 

[Zarathustra[H]]

Nah, yer not quite getting it. GIS a 2 pass radiator. You're cutting the rad in half essentially. Flow will then not go thru the traditional cold side, it would just flow thru the shortest least resistant path, all thru the hot side. Btw, I'm guessing the nexxos is a tradition 2 pass rad. If its a 2 pass like HWL... can of worms.

I'm not quite sure what some of your abbreviations are, but these radiators have no prescribed flow direction, thus by design there is no side intended to be a hot or a cold side, so both sides of the U should be the same.

 

[Tsumi]

The water will not all flow through the path of least resistance. It will be a ratio defined by inverses. Think of the equation for calculating the resistance of parallel resistors.

The only "greater resistance" one half will have over the other is the fact that the vent port is located off the center axis of the rad. I'm betting that difference will effectively be within the margin of error if you were to actually test it.

Furthermore, the flow resistance offered by the radiator will be half of what it is in the normal U configuration. Instead of forcing water through 8 channels for 720mm with a sharp 180° turn in the middle, you'd be forcing water through 16 channels for only 360mm and the flow direction would only ever change by 90°.

As long as you split your flow at the terminal end like your diagram shows so that your water is going into both sides of the "U", you'll effectively have an x-flow rad.

Put it way better than I could have.

For experimental purposes, put two identical funnels on the ports and see which one drops faster. The fluid dynamics of this setup will more closely mirror actual flow dynamics than putting the funnel on the vent port.

 

[Kardonxt]

Honestly I would just hook up a demo setup in a sink \ bathtub and verify. You will clearly loose some cooling efficiency, but this will at least let you know there is no unexpected restriction inside the rad. (I say bathtub because I used a sink to test a d5 before not realizing it would flow like a garden hose and drench my entire bathroom in 1/2 a second lol.)

 

[thesmokingman]

I'm not quite sure what some of your abbreviations are, but these radiators have no prescribed flow direction, thus by design there is no side intended to be a hot or a cold side, so both sides of the U should be the same.

You don't know what GIS means?? The cold vs hot side is dependent on your loops flow direction. And for the sake of debate it doesn't matter which side is which but you are oriented by the vent port. Martins already shown over a decade ago what happens to flow when a signle pump in a parallel setup goes down. The remaining pump saw 83% drop in power. That mean the flow went thru the path of least resistance.

Congrats, you cut your rad down to 1/3 size.

 

[Zarathustra[H]]

You don't know what GIS means?? The cold vs hot side is dependent on your loops flow direction. And for the sake of debate it doesn't matter which side is which but you are oriented by the vent port. Martins already shown over a decade ago what happens to flow when a signle pump in a parallel setup goes down. The remaining pump saw 83% drop in power. That mean the flow went thru the path of least resistance.

Congrats, you cut your rad down to 1/3 size.

At that point you are comparing back pressures.

In the parallel pump example the differences in back pressures is huge.

For the water flowing through the loop, you have the entire loop, plus all blocks and radiators. For flowing back through the parallel pump all you have to do is overcome the impeller of the dead pump.

It's no wonder 83% of the flow went back through the dead pump.

This example is very different. You have two flow options, one through either side of the radiator. One has slightly less resistance that. The other depending on the offset of the vent port, but in the grand scheme of things that will have a comparatively small impact on the resistance, because whichever path is chosen you still have the entirety of the loop behind that port.

When you have small differences in resistance based on the route taken, you also have smaller differences in the amount of flow between the routes.

If the resistance were 50-50, you get a 50-50 split of flow. here the resistance difference is going to be small. Let's call it 52-48. You are going to have a slightly higher amount of flow in the lower resistance path, but it's not going to be anywhere near that 83% figure you are quoting.

The split in flow between the two paths is going to be roughly equivalent to the split in resistance.

 

[thesmokingman]

I agree we won't know, er you won't know the split until you test it. However I wouldn't pull out a random number.

 

[Zarathustra[H]]

I agree we won't know, er you won't know the split until you test it. However I wouldn't pull out a random number.

Yep, I was just using the number for sake of example. I won't know for sure until I test the configuration.

I expect the port position to have a relatively small contribution to flow difference though. At least compared to your pump example.

 

[Tsumi]

Yep, I was just using the number for sake of example. I won't know for sure until I test the configuration.

I expect the port position to have a relatively small contribution to flow difference though. At least compared to your pump example.

I would expect the same as well. Numerous testing has shown that it takes several 90 degree elbows to match up to the resistance of a CPU block, and the 90 in a radiator is somewhat less extreme. My bet is less than 10% difference between the two sides (45/55 split or less).

 

[DWolvin]

Following to see the results... I'm on the 'negligible difference between the sides' (10-15%) group but will be Mythbusters happy to be incorrect.

 

[Zarathustra[H]]

Following to see the results... I'm on the 'negligible difference between the sides' (10-15%) group but will be Mythbusters happy to be incorrect.

Before I decide on my routing I am going to hook the radiator up to a faucet using a garden hose to 3/8" Barb adapter and flow water through it. (I hope residential water pressure is sufficient to simulate a D5 )

Then I am going to run tubes into two identical vessels and see if they fill at significantly different rates.

The flow will be in the reverse direction of my image above though, in the vent port, and out the two normal flow ports.

 

[DWolvin]

I'd think that would be the worst-case scenario, due to the asymmetrical plenum construction, but water doesn't compress to it may not matter much... Can't wait!

 

[VanGoghComplex]

I'm still betting you get performance that is comparable to a cross-flow rad.

Which is also to say that you'll get performance that is comparable to a U-flow rad hooked up normally.

 

[BrotherMichigan]

I'm still betting you get performance that is comparable to a cross-flow rad.

Which is also to say that you'll get performance that is comparable to a U-flow rad hooked up normally.

Depends on the rad. The HWLabs Nemesis X-flow rads are notably worse compared to the U-flow versions in reviews.

 

[lobstar]

I understand that would result in slower flow through the radiator, but I'm not convinced that is a terrible thing.

My understanding is that it would only use roughly half of the usual paths through the radiator and result in lower cooling capacity. I also inquired about this with my monstas to alpha before purchase. I believe they have a very similar core setup.

 

[VanGoghComplex]

My understanding is that it would only use roughly half of the usual paths through the radiator and result in lower cooling capacity. I also inquired about this with my monstas to alpha before purchase. I believe they have a very similar core setup.

If you only hooked up one of the two normal end terminals, yes - but if you Y them and hook them both up then the flow will split across all paths.

 

[lobstar]

If you only hooked up one of the two normal end terminals, yes - but if you Y them and hook them both up then the flow will split across all paths.

You are forcing a possible flow restriction and each 'source' of water coming into the radiator still only gets access to half as much surface area and thus less cooling. I'm sure it will be just fine and won't hurt anything just not optimal for cooling performance. Also, I'm not entirely sure what would happen to cooling performance for the portion of a loop that had a pressure differential between the inlets. Would it create suction in the second loop or would it cause less flow due to the end-cap design having the 'outlet' in your case offset from the inlets in distance?

 

[VanGoghComplex]

You are forcing a possible flow restriction and each 'source' of water coming into the radiator still only gets access to half as much surface area and thus less cooling. I'm sure it will be just fine and won't hurt anything just not optimal for cooling performance. Also, I'm not entirely sure what would happen to cooling performance for the portion of a loop that had a pressure differential between the inlets. Would it create suction in the second loop or would it cause less flow due to the end-cap design having the 'outlet' in your case offset from the inlets in distance?

I don't think it'll have any of those problems. Sending water into both terminals on one end and out a single terminal on the other is essentially the same thing as a cross-flow rad. The flow through each side of the radiator might be slightly different, but I seriously doubt it'll be enough of a difference to make any impact on cooling performance.

This thread is making me want to borrow my brother's unused Monsta rad and set up a test loop. =)

 

[Zarathustra[H]]

You are forcing a possible flow restriction and each 'source' of water coming into the radiator still only gets access to half as much surface area and thus less cooling. I'm sure it will be just fine and won't hurt anything just not optimal for cooling performance. Also, I'm not entirely sure what would happen to cooling performance for the portion of a loop that had a pressure differential between the inlets. Would it create suction in the second loop or would it cause less flow due to the end-cap design having the 'outlet' in your case offset from the inlets in distance?

I don't think you understand how this works.

Let's flip it around maybe.

Input on the single port on the non traditional side, output on the two ports on the traditional side. The two ports on the traditional side are then joined to one tube either via a y connector, or by both being routed to the reservoir.

When the water enters in the one port, it will split, roughly half will go on the left side, and roughly half will go on the right side.

Yes, each side will be half the cooling capacity, but it will also see half the amount of water to cool, so it roughly evens out.

There will likely be some efficiency losses due to the slower flow (half the flow onneach side) but aim betting that is on the noise level and not significant to performance.

 

[Zarathustra[H]]

I got my garden hose (faucet) to 3/8 Barb adapter today, but I didn't have the time to remove the radiator for flow testing yet.

I'll get to it at some point this weekend.

 

[DWolvin]

insert 'well, we are waiting' gif here...

 

[Zarathustra[H]]

insert 'well, we are waiting' gif here...

Yeah, everything is taking a little longer than anticipated.

I forgot how difficult it was to shoehorn that radiator in the top of my 750D, and getting it out wasn't much easier.

I have it out and cleaned now though, so I'll run my faucet test after dinner, and post back here.

 

[Zarathustra[H]]

Alrighty.

I did the test, and boy were the results surprising.

Now first the part that didn't surprise me. They are really close. I'm not going to bother measuring the quantity, but I am going to make an educated guess that the difference is like a quarter cup in a gallon. If that visual estimation is accurate, we are talking 50.4% on one side, and 49.6% on the other.

Now for the surprising part.

The side that the vent port is skewed towards was the jug that wound up with LESS water. I did not expect that, and can not find anything in my setup which would have caused this. The radiator is laying flat, straight tube going in, no kinks, etc.

So, weird I guess.

Anyway, based on these results, I'd argue this is a pretty safe way to hook things up. The difference in flow between the two sides is absolutely minimal.

 

[VanGoghComplex]

Alrighty.

I did the test, and boy were the results surprising.

View attachment 222269

View attachment 222270

Now first the part that didn't surprise me. They are really close. I'm not going to bother measuring the quantity, but I am going to make an educated guess that the difference is like a quarter cup in a gallon. If that visual estimation is accurate, we are talking 50.4% on one side, and 49.6% on the other.

Now for the surprising part.

The side that the vent port is skewed towards was the jug that wound up with LESS water. I did not expect that, and can not find anything in my setup which would have caused this. The radiator is laying flat, straight tube going in, no kinks, etc.

So, weird I guess.

Anyway, based on these results, I'd argue this is a pretty safe way to hook things up. The difference in flow between the two sides is absolutely minimal.

*goofy touchdown dance*

 

[Zarathustra[H]]

In preparation for this build I bought one of the EK Dual D5 pump tops, but I think I have changed my mind.

I'm leaning towards getting two separate D5 pump heads, and going with my other crazy idea instead. Not decided yet. Routing things the way I want them is complicated due to the Heatkiller IV having the inlet port on the right, and outlet on the left. I have worked on lots of tubing layouts and every single one I come up with results in crossed tubes on the CPU block.

The fact that the top and bottom screw holes have different spacings means you cant rotate it to change the direction of inlet and outlet. :/

 

[Zarathustra[H]]

Doh, slipped and dropped the 45mm 420 radiator from about 4ft onto a cardboard box on the cement basement floor.

Flowed some more water through it, and luckily it looks like it survived.

Going to have to give it a thorough leak test after assembly just to make sure.

 

[DWolvin]

Oof, that would suck to break it now.