What about minor losses in centrifugal pump sizing and system curve?

[Stephen.W]

I’m trying to size a centrifugal pump for a closed-loop cooling system, and I keep hitting a wall when it comes to the system curve. My calculated friction losses seem okay, but I’m genuinely unsure how to properly account for the minor losses from all the elbows and valves without overcomplicating the model.

[Mia.S]

For me the stubborn part was the minor losses. I ended up bundling all elbows and valves into a single effective K and using hL = K * v^2 /(2g). I list the fittings, pick rough K values from a table, multiply by the velocity in the relevant pipe section, and add that to the friction loss. It keeps the model simple and you can adjust K_total if you have flow data. The catch is the velocity: in a loop with different pipe diameters you need to pick a representative v for the section where those fittings sit, or split the model into zones. Also be careful you don’t double count losses across a reducer and elbows in the same stretch.

[Elizabeth.S]

I tried to go the full-lookup route once with every elbow and valve and it ended up overfitting. The system curve kept moving around as I changed K for one fitting. It taught me that minor losses are more of a uncertainty band than a precise term in practice.

[Sofia_L]

Concrete action I took: run a quick test with a clamp-on flow meter to get actual head loss at a known flow, then back out an empirical K_total that fits. It wasn’t perfect, but it gave a sanity check that the K_total wasn’t enormous.

[LoganWC]

One question: do you have a sense if the real bottleneck is the pump curve not intersecting your system curve, rather than the minor losses? Sometimes the whole exercise hides the fact the pump’s shutoff head or best efficiency point is far away from your target flow.