A centrifugal pump on a coolant system was making a gravel-rattling noise after six months. The pump was sized for 500 L/min at 3 bar head. The NPSH available was calculated at 6 m. The pump’s NPSH required was 3 m. On paper, there was 3 m of margin. The pump still cavitated. The customer thought the pump was defective. It wasn’t. The NPSH calculation was missing something.

The NPSH number that lies

NPSH required (NPSHr) is measured at the pump suction flange with clean, cold water. It’s the head at which the pump’s output drops by 3% (the NPSH3 point). But cavitation starts before that point. The noise — the gravel-rattling sound — starts at about 1.3x NPSHr. The actual cavitation-free operation point is NPSHa > 1.3 x NPSHr. The 3 m margin in the customer’s calculation was using NPSHr = 3 m. The real requirement was 3.9 m. The NPSHa was 6 m. That leaves 2.1 m margin. On paper it should work. But the customer’s fluid wasn’t cold water.

The coolant was a water-glycol mix at 60°C. Glycol raises the vapor pressure and lowers the specific gravity. The effective NPSHa drops by about 1 m at 60°C with 30% glycol mix. The real NPSHa was 5 m. The required NPSHa was 3.9 m. The margin was 1.1 m. That’s tight. And the suction pipe had a 90-degree elbow right at the pump inlet. That elbow creates turbulence at the eye. The effective NPSHr goes up by 0.5 m at the pump. The margin drops to 0.6 m. The pump cavitated.

What fixed it

The suction elbow was moved 5 pipe diameters upstream of the pump inlet. A straight section of pipe between the elbow and the pump lets the flow stabilize. The turbulence at the eye dropped. The effective NPSHr dropped back to 3 m. The margin increased to 2 m. The cavitation noise stopped.

Also, the glycol concentration was reduced from 30% to 15%. The system only needed 15% glycol for freeze protection (the plant was indoors). The extra glycol was unnecessary. The vapor pressure dropped. The NPSHa increased by 0.3 m. The margin grew to 2.3 m. The pump ran quietly.

The customer also installed a pressure gauge at the suction flange. The gauge reads the actual suction pressure. If the pressure drops below the NPSHa equivalent, the pump is cavitating. The gauge gives an early warning before the noise starts. It’s a $50 gauge that prevents a $5000 pump replacement.

The NPSH rule that works

Size NPSHa at 1.5x NPSHr minimum. For glycol or hot fluid, use 2x. For suction elbows within 5 pipe diameters of the pump, add 1 m to NPSHr. These rules sound conservative but they prevent the gravel-rattle that comes back every time the suction filter gets slightly clogged. The margin absorbs the unknowns — dirty filters, warmer fluid, slight elevation changes.

If the pump is already cavitating, check these in order: suction filter (is it clogged?), suction pipe (any air leaks?), fluid temperature (is it hotter than calculated?), suction elbow (too close to pump?). Don’t start by replacing the pump. Start with the cheap fixes.

NPSHr is measured on cold water with a straight suction. Real-world fluid, temperature, and elbow geometry add up. Size at 1.5x NPSHr (2x for glycol/hot fluid), keep 5 pipe diameters of straight suction, and put a gauge on the suction. The whine isn’t a defective pump — it’s a system that didn’t leave enough margin.