Net Positive Suction Head Available (NPSHa)

Because the vapour pressure Pv is an absolute pressure, and the difference (Ps − Pv) has to be evaluated on the same basis. Gauge pressure is measured relative to local atmospheric pressure, which itself varies with altitude and weather — so a gauge pressure of 0 kPa means 101.3 kPa absolute at sea level, but only 89.9 kPa at 1,000 m elevation. The physics of cavitation is governed by absolute pressure, because the liquid flashes when its absolute pressure falls to the vapour pressure, not when it falls to some gauge value. Always convert gauge readings to absolute by adding the local atmospheric pressure before using the NPSHa formula, and always use absolute pressures for both Ps and Pv.

API 610 recommends a minimum NPSH margin of 1 m for hydrocarbon pumps handling liquids near their boiling point. In practice, most operators use 1.5–3 m to allow for the uncertainty in the calculation (especially the friction-loss estimate), for wear of the impeller and clearances, for operation away from the duty flow, and for the fact that NPSHr rises steeply with flow — a pump that has a 1 m margin at its duty point may have a negative margin at 120 % of duty flow. For clean cold water in a well-designed installation, a 1 m margin is often acceptable; for hot liquids, volatile hydrocarbons, or viscous slurries, 3 m is a much safer target. The cost of an extra metre of suction head is trivial compared to the cost of a cavitating pump.

Because the vapour pressure of a liquid rises extremely steeply with temperature — much faster than the pressure head available. For water, Pv goes from 2.34 kPa at 20 °C to 47.4 kPa at 80 °C — a factor of 20 — while the density changes by less than 3 %. So the (Ps − Pv) term in the formula shrinks dramatically as the temperature rises, and the pressure head (Ps − Pv)/(ρg) with it. At 20 °C, with Ps = 101.3 kPa, the pressure head is about 10.1 m; at 80 °C it is only 5.66 m — a loss of 4.45 m of NPSHa from the temperature change alone. This is why hot-liquid pumps are always designed with generous margins, and why the cold-water test of a pump that will eventually run hot is not a valid check of its suction performance.

Add the distributed friction (the pipe itself, from the Darcy-Weisbach equation or from a friction-loss table) to the minor losses (every bend, elbow, tee, reducer, valve, strainer, and the pipe entrance). Distributed friction scales with the square of the flow velocity and with the pipe length, and inversely with the pipe diameter. Minor losses are usually expressed as velocity heads (K·v²/2g) with K values tabulated for each fitting — for example K ≈ 0.5 for a sharp-edged entrance, 0.9 for a standard elbow, 0.3 for an open gate valve, 5–10 for a clean strainer, and 15–20 for a partly fouled strainer. Add up all the K values, multiply by v²/2g, and add to the pipe friction. When in doubt, be conservative — over-estimate hf rather than under-estimate it, because a fouled strainer or a partially closed valve can multiply the losses in service.

Yes, in several ways, though some are cheaper than others. The easiest is usually to raise the liquid level in the suction tank — even a few tens of centimetres of additional level may be enough. Next easiest is to reduce the friction loss: replace the suction strainer with a larger or coarser one, clean or replace a fouled strainer, remove an unnecessary valve or fitting, or increase the diameter of a section of suction pipe. On an existing pump, it may be possible to slow the pump down or trim the impeller to reduce the duty flow and therefore the NPSHr — this is a common fix for a marginal pump. The most expensive option is to replace the pump with a lower-NPSHr model, but sometimes it is the only practical choice if the installation cannot be modified.

Because NPSHr is a strong function of the flow rate. As the flow increases, the liquid enters the impeller at a higher velocity and the local static pressure at the impeller eye falls further — so more suction head is needed to keep the liquid above its vapour pressure. On a typical centrifugal pump curve, NPSHr rises roughly as the square of the flow rate, which means that a pump that runs comfortably at its best-efficiency point may cavitate badly if operated at 120 % of that flow. This is why the NPSHr must always be read off the pump curve at the actual operating flow, not at the rated flow, and why the NPSHa must be checked against the worst-case flow the pump will ever see. It is also why a throttled-down pump has a lower NPSHr than the same pump running wide open.

In the short term, cavitation produces a characteristic rattling or crackling noise, vibration, and a drop in head and flow. In the longer term, the collapse of the vapour bubbles against the impeller surface pits the metal, gradually eroding the impeller vanes, the volute, and sometimes the casing. A severely cavitating pump can destroy its impeller in a few weeks, and the resulting vibration can damage bearings and seals. In hot hydrocarbon service, cavitation can also be a safety hazard, because the vapour released may be flammable. This is why NPSH is taken so seriously in pump design: not only does cavitation destroy the pump, it often does so silently and gradually, so the failure is discovered only after significant damage has occurred.