Not sure how many of these dogie has, but it'd be a cool test to do anyway. You wouldn't even need to use that rad, you could just run three of the 360mm ones he sells in series, and maybe a couple of the pumps to generate sufficient head.
*Edit: That's a minimum flow rate to keep the inlet water temperature of the last unit at a reasonable (say 65C with 35C inlet water) temperature. Depending on the block design you might need a higher rate to actually get 1kW per device from the block into the water.
Lets try some maths:
flowrate = Q / ((heat capacity)* density *(Tout - Tin))
Q = 5KW
Heat capacity of water = 4.18 kJ/kg/K
Density = ~1000kg/m^3
Tout max = 65C to keep chips < 75C [pretty optimistic heat exchanger]
Tin = 30C [depends on ambient, I wouldn't be getting mine below 40C]
flowrate = ? m^3/sflowrate = 5 / (4.18 * 1000 * 35)
flowrate = 0.0000342 m/3^s = 0.0341 L/s = 122 L/h
With a 6mm internal diameter [forced to this size by the heatsink channels]...
A= 0.0000283 m^2
flowrate = A*v
v = velocity m/s
0.0000342 = 0.0000283 * v
Therefore v would be 1.21 m/s.
Assumes perfect heat exchanger, which its not so we lose effective deltaT. Do remember though the size of the pump you're going to need to achieve 120 L/h with that level of restriction.
Assume 40% effectiveness due to single flow.
New v = 1.21/0.4 = 3.0 m/s
Minor losses:
v of 3.0 m/s
K of 40 [conservative, if you think of the number of times we have 180 degree turns in the 5 blocks + the radiator]
head loss (m) = 18.4m
Major losses:
Assume 20m of internal flow path = L
D = 0.006m
V = 3 m/s
Major losses = 30.6m
Total losses = 49.0m.
So if you can generate 50m of pump head, you can move the fluid.
