How do I size a centrifugal pump for a closed-loop cooling system?

[Jonathan48]

I’m trying to size a centrifugal pump for a closed-loop cooling system, but I’m getting stuck on the system curve. My calculated head seems way higher than what the pump charts show is available for the flow rate I need, even after accounting for pipe friction and fittings. I’m wondering if my assumption about the equivalent length for the valves is throwing everything off.

[GraceR]

I've run into this. The valves can wreck your head prediction if you treat them like an extra pipe length. The trick is to use a loss coefficient instead of an equivalent length guess. h_loss = K * V^2 /(2 g). Get K from the valve vendor or from a curve, and use the actual velocity in the main line. Then add up all the losses with the right flow, not just rough lengths. Once I did that, the curve moved a lot and the mismatch eased, but it still wasn’t perfect.

[Alexander18]

I've done the opposite mistake once: I assumed a valve added a certain length, and I overestimated it; with fittings I double-counted. The numbers looked nonsense until I physically measured flow with a handheld meter and checked what the valve was actually doing. It turned out the valve was far more open than I modeled, and the loss dropped by about half. The system curve changed a lot once I corrected that.

[AuroraJ]

Another angle: maybe the problem isn't the valve losses but the pipe diameter and the Reynolds number. If the flow is high, the friction factor f depends on roughness; I once used a static estimate and got a steeper curve than reality. Running a proper friction factor from Colebrook and redoing the pipe losses brought things closer, but it still wasn't perfect.

[MilaR]

One question: are you sure you're including the static head correctly, like elevation change in the loop? If the loop climbs a few meters, that adds to the head you need, and if your model assumes zero static head you’ll see a big mismatch.