[SOLVED] Which radiator should I put first ?

[tek3195]

I have a 240mm AIO with an additional 120mm radiator added to the loop. The 240 is in front as intake and the 120 is up top as exhaust. Each time I think of which direction to run flow, the time before made more sense. Should I have it flow from pump to 120 first exhausting some heat before going to 240 or the other way around ?

i7-9700 @4.5GHz with power limit set to 300W
Asrock B365M Phantom Gaming 4
4x8GB Team ELITE @3000MHz
Samsung 970 EVO M.2 SSD 500GB
SK Hynix 2.5" SSD 1TB
Cooler Master q300l
Alphacool Eisbaer 240 black AIO
Alphacool NexXxoS ST30 120mm Radiator
5x Arctic Bionix p120
Rosewill Photon 550
Debian Bullseye with windows portable on usb drive

 

[Ralston18]

Update your post to include full system hardware specs and OS information.

What options are shown in the case, cooler, fans, etc. documentation: User Guides & Manuals.

"I have a 240mm AIO with an additional 120mm radiator added to the loop. The 240 is in front as intake and the 120 is up top as exhaust. Each time I think of which direction to run flow, the time before made more sense. Should I have it flow from pump to 120 first exhausting some heat before going to 240 or the other way around ? "

Could be none of the above.

Sketch out the build and assemble per the documentation.

Doing otherwise may end badly and/or void warranties.

Just my thoughts on the matter....

 

[Karadjgne]

You want the least amount of travel, bends, crossing. So which you do first will depend on the actual direction of flow and ease of flow.

With a pump in front low, it's generally easier to go from pump to gpu, rear rad, cpu, top rad, front rad, reservoir, but can always swap rear/cpu around depending on whether the cpu block is left intake or right intake or top/bottom intakes etc.

Ideally you'd be looking to make a clean circle out of the loop, but things can change depending on exactly what rads are used, like differences with traditional U flow, or an X flow (flow through) etc.

Draw it out first, plan it out according to dry fitting all the rads, making note of any clearance possible issues that might require turning a rad 180° and totally changing its input/output fitting placements.

There's really no right way or wrong way to planning a loop flow direction or placements, it's more of a hard way or easier way. You can have a complete spiderweb of criss-crossing pipes and still have correct flow, it just makes it harder to plan, harder to fit, harder on the pump etc.

 

[tek3195]

Update your post to include full system hardware specs and OS information.

What options are shown in the case, cooler, fans, etc. documentation: User Guides & Manuals.

"I have a 240mm AIO with an additional 120mm radiator added to the loop. The 240 is in front as intake and the 120 is up top as exhaust. Each time I think of which direction to run flow, the time before made more sense. Should I have it flow from pump to 120 first exhausting some heat before going to 240 or the other way around ? "

Could be none of the above.

Sketch out the build and assemble per the documentation.

Doing otherwise may end badly and/or void warranties.

Just my thoughts on the matter....

Updated post.
My build isn't actually supported by case manufacturer and Alphacool simply makes the option to customize available, no documentation I've found for custom options.

 

[tek3195]

You want the least amount of travel, bends, crossing. So which you do first will depend on the actual direction of flow and ease of flow.

With a pump in front low, it's generally easier to go from pump to gpu, rear rad, cpu, top rad, front rad, reservoir, but can always swap rear/cpu around depending on whether the cpu block is left intake or right intake or top/bottom intakes etc.

Ideally you'd be looking to make a clean circle out of the loop, but things can change depending on exactly what rads are used, like differences with traditional U flow, or an X flow (flow through) etc.

Draw it out first, plan it out according to dry fitting all the rads, making note of any clearance possible issues that might require turning a rad 180° and totally changing its input/output fitting placements.

There's really no right way or wrong way to planning a loop flow direction or placements, it's more of a hard way or easier way. You can have a complete spiderweb of criss-crossing pipes and still have correct flow, it just makes it harder to plan, harder to fit, harder on the pump etc.

Mine is a bit more simple since it consists of Alphacool pump/res combo two radiators and soft tubing. It's just their standard aio that I changed tubes to clear, got rid of big ugly quick disconnects and added second rad. Only 3 tubes, pump>rad, rad>rad and rad>pump.

 

[Paperdoc]

Karadjgne's post above is important respecting smooth loops and flows. IF you can achieve that no matter which is the sequence of rads, then consider this. This comes out of what's known in process engineering as Countercurrent Flow. The basis is that the rate of heat transfer across any heat exchanger depends in part on the temperature DIFFERENCE between the hot and cool media - in this case, the liquid in the loop and the air blowing through the rad. You want to maximize that difference on each rad.

Case 1: Fluid flows from hot CPU cooler block to the top (exhaust fans) rad first, then to the front (intake fans), and back to the reservoir/pump. In the top rad the fluid is hotter, and the air blowing though the rad is pre-heated from the mobo components; these two factor affect the magnitude of the temp difference. At the front rad, the fluid is already slightly cooled from the first rad, and the cooling air is as cool as possible because it comes from outside.

Case 2: Reverse the rads in the sequence. Now the fluid at its hottest flows to the front rad first and is cooled by the coolest incoming air. That's the max temp difference you can get at the front rad. After that the fluid goes to the top rad at a lower temp than in Case 1, BUT the air flow though the rad is still pre-heated from the case interior. So the temp difference here is LESS than in Case 1.

Normally it turns out that in Case 2 above with more heat removed at the front, the heat removal at the top is much less than in Case 1, and hence the TOTAL heat removal rate of the two rads is LESS than Case 1. So it is better to have them arranged so that you get a decent temp difference at each rad, rather than trying to maximize the difference at only one.

View it this way: you are trying to move something from one medium to another using an extraction medium, with two transfer devices in series. The "something" is heat, the medium with MORE of "something" is the fluid in the loop, and the extraction medium is air which has LESS of the "something" in it. The guideline is the extracting medium should FLOW in the opposite direction from the flow of the material with more of "something". So the outside coolest air should flow into the series system at the end where there is the least temperature, and then on to the second stage that operates at higher temperature but still removes heat because that air is still cooler than the liquid.

 

[rubix_1011]

Yep, the primary influence here being the volume of ambient (cool) air hitting the radiators, which would be the only real difference...the rest being equal.

There aren't any real 'hot' components in a water or liquid cooling loop, but for the sake of semantics and understanding where load is being exchanged from the CPU or GPU, you can consider those 'hot' components to be the points at which thermal energy is introduced/exchanged into the liquid of the system. Radiators being the 'cool' points, or exchange of thermal energy to the surrounding air - whether that be inside or outside of the case.

Liquid cooling is dependent on the following factors and will be better or worse depending on changing any combinations of each:

1) coolant flow rate - pump head and flowrate - loop restriction is a direct deterrent to overall flow. Every loop has restriction, so every loop will operate at a lower speed than possible that any given pump is advertised. AIO pumps move very low volumes of water, while some dedicated watercooling pumps can move many liters/gallons per minute. Restriction and flow are impacted by tubing diameter, restrictive bends (90 degree angles, etc) as well as block restriction. Even straight tubing offers some restriction in flow over the distance of several meters.

2) Thermal Conductivity vs. Specific Heat of metals - this applies to both cooling blocks on CPU or GPU as well as metals used in radiators. Ex; using aluminum or copper makes a difference

3) radiator surface area - heat in watts produced by the CPU and GPU in a cooling loop need to be effectively removed by radiators. If you have 400w of output heat by CPU and GPU, but only have a 120mm or 240mm radiator which can dissipate between 150w and 300w, you still have an additional 100w being constantly added, which cannot be removed. This is where calculating loop thermal capacity is important. Just by 'having a radiator' does not mean you 'have enough radiator' space.

4) cooling fans - high airflow/high CFM fans look good on paper, but might not have the static pressure (in either mm/O2 or mm/HG) to effectively to force air through a radiator. Static pressure designed fans are meant for dense radiator fins or applications where the pressure of air being moved is more important than the actual CFM of that column of air.

You can improve any component of the cooling loop values above and improve the overall cooling ability of the loop. If you decrease 1, to remain effectively 'equal', something else will need to compensate for that value change. Ex: using lower speed fans might require more radiators to be used and possibly higher flow rate.

 

[tek3195]

Awesome info and delivery. @Paperdoc @Karadjgne @rubix_1011 ya'll have explained in a surprisingly easy to understand way. I wasn't really expecting an answer on that level. The same question in Alphacool forum yielded "it doesn't really matter" from an alphacool staff member. I knew it mattered and am glad I asked it here. I did get the fans right by swapping stock for Arctic Bionix P120 which from what I read have very good static pressure tuned for radiator and other restrictive applications. This is first PC build, so learning a lot. Lifetime of fabrication experience with PC experience coming in at zilch. I see computer building is a bit different than a tube chassis race car after all. How can I mark three answers as best ? Can't single one out, they all go together.

 

[Paperdoc]

Thanks, your comments are appreciated. I'll just add clarification on the pressure rating of fans, which rubix_1011 has rightly pointed out is important when your application is a radiator or a finned heatsink.

The way I understand fan specs is this. You'll see these two items (plus others) spec'd on most fans: Air Flow and Pressure. For any fan if you sketch a curve of the actual air flow produced versus the backpressure it faces (resistance to air flow), that line is VERY roughly a straight one. You get the specified MAX air flow for zero backpressure - no impediment to air flow. As air flow resistance is increased (measured as the air pressure above ambient right after the fan discharge, in mm of water column) the actual flow produced decreases, until you get to that Pressure rating. At that pressure (flow resistance) and beyond there is NO appreciable air flow. Of course, this measurement is done at the fan's max speed. For identical motors, the graph may be different depending on details of the fan blade design. Unlike what I thought at one time, you do NOT get the rated max Air Flow at the same time as the max air Pressure.

So in comparing fans, if you sketch out on the same graph the lines for each fan - each one a straight line from highest flow at zero backpressure to zero flow at max rated backpressure - you can see how they will perform under various conditions. The fans sold as "Pressure" fans for rads etc. normally deliver less air flow at zero backpressure, but are much better at delivering air flow when there is some backpressure. The "Air Flow" optimized fans are great for flow at very low backpressures (like in a well-organized case), but cannot generate much flow at all when trying to push though a rad or a finned heatsink. Now, all of that you can do for the fans running at max speed. You almost never will see comparable specs for flow and pressure limits when the fan is at reduced speed. But that's OK. As long as the fan can deliver more than you need at max speed, any automatic fan speed control (which really is based on maintaining the TEMPERATURE of the heat source component) can reduce the fan speed to deliver whatever air flow IS required at that moment.

 

[rubix_1011]

Also, for the most part, the order of components in the loop are not important, so whether your CPU block or GPU block are 'before' or 'after' 1 radiator or both radiators, it really makes little difference.

The caveat being - this also depends on which of these radiators have fresh/cool air and which are using air from within the case, which still should be minimal difference unless you have a very poor airflow case. If airflow is very good, most of the discussions in this thread are negligible.

Coolant will operate at a very consistent equilibrium once the system is at operating temp (i.e. the coolant has reached a stable working temperature under sustained load). Again, this changes if you alter fan speed on radiators (ex: if set to a PWM curve) or if your pump is set to PWM and ramps up and down (which please don't tell me if you do, I don't want to know). My advice would be to leave the pump at 100% or some other fixed setting and only alter radiator fans via PWM if you want.

Also, coolant temps are not the same as die temps reported by software such as HWinfo, CoreTemp or RealTemp, etc, which is exactly the reason why liquid cooling pumps being set to CPU die reported temps is a poor idea as they can change rapidly within the same second...what's the point of ramping a pump up and down when the coolant is still only able to absorb heat based on its specific heat. http://www.kentchemistry.com/links/Energy/SpecificHeat.htm
also
https://www.engineeringtoolbox.com/heat-work-energy-d_292.html

This shows that water (l = liquid) has a much higher specific heat capacity than many other substances at 25C. The higher the value on a specific heat index means it takes more thermal energy to raise the overall temperature of the set unit of of the substance meaning that water can absorb a tremendous amount of energy per gram before the measured temperature raises by 1C.

Again, the entire argument against having a liquid cooling pump set to PWM control.

 

[USAFRet]

which is exactly the reason why liquid cooling pumps being set to CPU die reported temps is a poor idea

QFT.

This needs to be repeated again and again and again.
Coolant needs to flow at the same rate....let the fans ramp up and down.

 

[tek3195]

Also, for the most part, the order of components in the loop are not important, so whether your CPU block or GPU block are 'before' or 'after' 1 radiator or both radiators, it really makes little difference.

The caveat being - this also depends on which of these radiators have fresh/cool air and which are using air from within the case, which still should be minimal difference unless you have a very poor airflow case. If airflow is very good, most of the discussions in this thread are negligible.

Coolant will operate at a very consistent equilibrium once the system is at operating temp (i.e. the coolant has reached a stable working temperature under sustained load). Again, this changes if you alter fan speed on radiators (ex: if set to a PWM curve) or if your pump is set to PWM and ramps up and down (which please don't tell me if you do, I don't want to know). My advice would be to leave the pump at 100% or some other fixed setting and only alter radiator fans via PWM if you want.

Also, coolant temps are not the same as die temps reported by software such as HWinfo, CoreTemp or RealTemp, etc, which is exactly the reason why liquid cooling pumps being set to CPU die reported temps is a poor idea as they can change rapidly within the same second...what's the point of ramping a pump up and down when the coolant is still only able to absorb heat based on its specific heat. http://www.kentchemistry.com/links/Energy/SpecificHeat.htm
also
https://www.engineeringtoolbox.com/heat-work-energy-d_292.html

This shows that water (l = liquid) has a much higher specific heat capacity than many other substances at 25C. The higher the value on a specific heat index means it takes more thermal energy to raise the overall temperature of the set unit of of the substance meaning that water can absorb a tremendous amount of energy per gram before the measured temperature raises by 1C.

Again, the entire argument against having a liquid cooling pump set to PWM control.

Pump set to PWM I am not guilty of yet, and won't be now that I've read this. I was going to, so thanks a ton for intervening. This thread has provided a wealth of knowledge for a first time builder. I was already seeing some mistakes I had made, such as customizing an aio having only 5/16" ID tubing, having my mind set on PWM for pump and a few others. This info will help make the best of what I have and go a long way on future builds or modification on this one. I don't imagine I'll change much of this one though, had thought about larger tubes but would need different pump to see any real benefit from it. Which would mean adding a reservoir and before you know it having nothing left from original system but a couple copper rads. That would be throwing good money after bad the way I see it. I believe the future holds a damn fine hard tube custom loop for me. Again, thanks and I really appreciate the info and links provided by all.

 

[Karadjgne]

Ehh, guilty of pwm ddc. But was done for a reason. I wanted a silent pc, and mITX is highly restricted for dual radiators. So instead of going with a slower, lower pressure, lower flow alphacool LT ddc, I went with the EKwb 3.2 pwm ddc, and turned down the rpms. Generally running 1800-2400 rpm, which is higher than the LT, but can run 4400 if I decide to stress. The 3.2 also came with an included heatsink, which I'm a fan of, looks cool and matches the build aesthetics nicely.

So yes, a method to the madness. But dual rads in a shoebox was definitely a lesson in madness anyway.

I like 10/16 for soft tubing. Still has the same flow as 10/13, but has a considerably thicker outer diameter, which is great at preventing kinks, doesn't suffer nearly as much sag when warm coolant softens up the tubing rigidity and has a lot more 'meat' for connector jackets to grab onto.

One of my pet peeves is the fan to rad ratio. It's kinda important to match the fan up with the right rad, or vice-versa. All too often looks are a major player, and thats honestly a minor concern. If you put the wrong fan on the right rad, you get zip for performance. If you put the right fan on the wrong rad, same thing, zip. Rads are not equal. Aio rads are pretty much jack of all trades rads, so excel at nothing, pretty mediocre. But custom rads are very different. Some perform excellent with high speed/cfm fans and very poorly with low speed/cfm fans, and some rads are the opposite. All require good static pressure, but there is a big difference between a 1200rpm fan with 2.5mmH²0 and 50cfm and a 2500rpm fan with 2.5mmH²0 and 60cfm.

Watercooling can get expensive. More so if you do it right, which doesn't always agree with the budget. It's almost never a popularity contest with brands, that's a bad way to go about it, it's all about what's actually going to work within the stipulations you require.

Just stay away from chalky pastels for coolant. Sure they look fantastic, but those show cases are broken down and cleaned within a couple of hours. Not for long term use. They'll clog up pumps and blocks, put undue wear and tear on pumps pushing such a viscous liquid, generally just a disaster for every day use. Stick with distilled water, a little biocide and maybe some transparent dye.

 

[rubix_1011]

Yep, agree with the radiator/fan discussion.

But to clarify, nearly every AIO uses aluminum radiators, so take heed with copper and brass in a loop containing aluminum.

 

[tek3195]

Yep, agree with the radiator/fan discussion.

But to clarify, nearly every AIO uses aluminum radiators, so take heed with copper and brass in a loop containing aluminum.

That is one decision I feel good about, Alphacool AIO rads are all copper. Not sure how they compare with other aio rads performance wise, but the copper was the selling point for me.

 

[rubix_1011]

Ah, yes, I missed the Alphacool Eisbaer above - I tested that one...I liked it, not super overall, but does a lot of cool stuff, like being easily expandable for a 'custom' like loop.

Nicely done, sorry that I overlooked that - but good choices there. If I were new to liquid cooling but wanting to do more, that's likely how I would recommend anyone starting out.

 

[Paperdoc]

Thanks for Best Solution. This is a fine example that highlighting one post in a thread may not be entirely fair. I think all of the posts here are excellent, and bring a lot of viewpoints and hard info to the discussions.

 

[tek3195]

Thanks for Best Solution. This is a fine example that highlighting one post in a thread may not be entirely fair. I think all of the posts here are excellent, and bring a lot of viewpoints and hard info to the discussions.

I could not agree more. I said it after the first round of answers, they all go together. The answers, the points of view and the proven facts compliment one another in a way that I think should be stickied on the first page of the cooling section. It's not that uncommon of a question, I don't think. I opened this page probably a dozen times to mark a Best Solution and closed it again unable to decide. I finally just kinda pointed with my eyes closed because I wanted it marked as solved where others may see it and gain from it like I have. I've asked a few questions in the past two years since learning to use a PC(yeah got first PC at 50 yoa) and this is the most informative and in depth response I've seen yet.