Start with the cavitation principle
KSB defines cavitation as the development and sudden collapse of cavities, such as vapour bubbles, in a fluid flow. It distinguishes vapour cavitation from gas cavitation. Vapour cavitation develops when the static pressure in the fluid falls below the vapour pressure associated with the fluid's temperature, KSB adds that this takes place without external heat supply and that small nuclei, often microscopic vapour bubbles, are also a prerequisite. In gas cavitation, bubbles form because dissolved gases come out of solution as pressure falls below the saturation pressure of those gases. Because that saturation pressure is often higher than the vapour pressure, gas cavitation can occur at pressures above the vapour pressure.
KSB explains that as the fluid flows through the impeller, the static pressure falls relative to the pressure upstream of the impeller, especially at the inlet to the vane passage. How large that drop is depends on rotational speed, the density and viscosity of the fluid, the inlet geometry of the impeller, the operating point and the velocity profile of the approach flow. Several of these factors belong to the pump itself and its operating condition, which is why the pump's side of the NPSH comparison cannot be read off from the installation alone.
Two quantities to compare
KSB states that assessing the likelihood, extent and impact of cavitation requires comparing two values: the NPSH required by the pump, NPSHR, and the NPSH available in the system, NPSHA. They are not two names for one number. NPSHA describes what the installation provides, referred to the suction nozzle's centre, KSB's page presents the definition as formulas and notes that the head loss includes entry losses and pressure drops across valves and fittings. For pumps without a conventional suction nozzle, such as welded-in in-line pumps or submersible pumps with bellmouths, the location of the reference point must be defined and stated.
NPSHR is the pump's side of the comparison. KSB specifies that its defining sum must not fall below a minimum value for a given pump and application, and that if this condition is not met, cavitation cannot be ruled out. Crucially, an NPSHR value only means something together with its cavitation criterion. KSB lists several: incipient cavitation (NPSHi), a certain extent of the cavitation zone on the vanes, the start of head drop (NPSH0), and a cavitation-induced head drop of 3 per cent (NPSH3). Demonstrating NPSHi requires demanding and expensive testing, so it is commonly agreed that NPSHR is taken as NPSH3.
That last point deserves care. NPSH3 is a defined head-drop criterion, a 3 per cent drop in head, and the value is not a threshold below which cavitation is absent. KSB's own criteria list shows that bubbles can appear earlier (NPSHi). KSB also shows that these values vary with the operating point, so a single figure applies to a stated condition, not to the pump in general.
Why temperature belongs in the discussion
KSB notes that there is a relationship between vapour pressure and temperature, so the same suction pressure sits at a different distance from the vapour pressure when the liquid temperature changes. KSB says that to avoid cavitation, or to limit it to an acceptable level, the pressure upstream of the impeller must exceed the vapour pressure by a specified minimum margin. The pages do not give one margin valid for all liquids and pumps, the margin is tied to the pump, the application and the cavitation criterion chosen.
NPSHR is established largely through testing. KSB specifically notes the importance of tests when converting between speeds, similar pump sizes or fluids. A value from a different test condition should therefore not be silently treated as an installation-specific requirement.
Keep the limits of the explanation clear
This article explains what has to be compared. It does not calculate the available NPSH of any installation and does not specify a permitted operating point. KSB does describe symptoms such as increased noise, rough running and drops in head and efficiency as indications that cavitation is taking place, but it also says that material damage does not occur in every case of cavitating operation. The sources therefore do not support diagnosing cavitation from a single noise or vibration observation. Pump data, liquid conditions and an approved engineering assessment are needed. When recording or discussing a result, NPSHA and NPSHR should be written as the separate quantities they are, with the NPSHR criterion stated.