Theorytical Proper GPU and CPU Waterblock Liquid Cooling


Oct 15, 2016
I've seen customloop buids photos and they're dazzling, I also notice that some are pure aesthetic not overclocking oriented. My reason for choosing watercooling is to get faster task completion by dissipating components heat faster so CPU or GPU perform above their TDPs. I know because one hour overclock watercooling better than three hours stock air cooling, so articles reading time taken for choosing the best available pumps, waterblocks and radiators is worthwhile rether than deadline lost hours in air cooling.
I don't see why manufacturers would bother enlisted their waterblocks, radiators, or fittings with pressure restricting values because reviewer such martinsliquidlabs, thermalbench, or extremerig would done it for them by sending them samples.
My company plan to upgrade the network renderer to upcoming nVidia GTX 1080 Ti and make the running 780 Ti as emergency solution. I've doing the power consumption calculating, running custom loop on 8 graphic cards is cheaper than individualy hybrid cooling.

1. Pump Capability
2. Expected Pressure Drop Restriction Point
3. Cooling Power Parts

#1. Pump Capability
Manufactur's PQ diagram (P = Pressure in mH2O, Q = Flow Rate in Liter Per Minute), for example the Koolance PMP 500 is rated 4 mH2O static head pressure at 9 LPM or 2.3 GPM (9/3.8), I assumed that at 4 meter loop (tubes + joints + waterblocks + radiator) the pump is flowing about 9 liter water per minute. At 6 mH2O it flowing about 1 GPM, you should verify yourself by including a digital manometer for pressure reading and King 7520 (Valved) for flow rate reading as already done by Martinsliquidlab . The test shows that at 6.6 mH2O flowing 1 GPM @ 12V. I figure that manufacturer's specifictions have been intended for free flow or without restriction variable values of waterblock's microfins and radiators pipe's, joints and fittings at the loop, which also included a near zero pressure drop.
1a. EK made dual D5 pump top, according to spec, it double the pressure head, does anyone ever done coupling PMP 500? I don't know if koolance made pump top for dual PMP 500 or, are there other top that will work for those two?
1b. What about this dual loop setup: pump1 > top1 > reservoir1 + reservoir coupler (with inlet 1 & 2) < reservoir2 < top2 < pump2. Orientation horizontal, two radiators, two waterblocks, pump1 loop: top1 outlet ~ waterblock1 ~ rad1 ~ res coupler inlet1, top2 outlet ~ waterblock2 ~ rad2 ~ res coupler inlet2.

#2. Expected Pressure Drop At Restriction Point
I reckon even the positioning of tubes and fittings selection will add restriction in loop, while the flow rate spread will be lesser as the loop gets longer devided throughout variable waterblock's microfins and radiators pipe's.
Question: (R = restriction value to PQ)
2a. R of different tube/pipe inner diameter (10mm, 12mm, 13mm)?
2b. R of flow tube/pipe (i)orientation (vertical upward, vertical downward, horizontal flow) and (ii) material (soft, stiff)?
2c. R of angled fittings (90 deg, 45 deg, multiple 45 deg)?
2d. R of coolant color additive ( does dye makes coolant less fluid or aqueous)?
2e. What waterblock you currently in use and its tested pressure drop?

#3 Cooling Power Parts
3a. What's your currently in use graphic card(s) and CPU(s) waterblock and its recorded temperature? is higher flow rate equal cooler temperature?
3b. Is there any safe or minimum flow rate limit, in order to find sweet spot between overclocked parts to power consumption.