Pump Cavitation: Field Diagnosis & Decision Tree

The 30-Second Triage

If a centrifugal pump is making noise, vibrating, or losing head, walk through these four checks before calling it cavitation. Most of what people label "cavitation" is actually NPSH problems, recirculation, or air ingress.

Rule of thumb: True classic cavitation sounds like marbles or gravel in the casing, not a hiss or a whine. If it sounds like a high-frequency hiss, it's almost always discharge recirculation.

The Four Cavitation Modes

Mode	Sound	Vibration Signature	Where Damage Appears	Root Cause
Classic NPSH cavitation	Marbles / gravel	Broadband, 0.5-15 kHz	Impeller eye / suction side of vane	NPSHa < NPSHr
Suction recirculation	Loud crackling, intermittent	Low frequency surges, 1-5 Hz	Pressure side of inlet vane tip	Operating well below BEP (typically <50%)
Discharge recirculation	High-pitched hiss / whine	Broadband + 1x vane pass	Pressure side of vane near discharge	Operating well below BEP, oversized pump
Vapor binding (air ingress)	Erratic flow, pump "loses prime"	Random, no consistent pattern	Usually no impeller damage	Suction line leak, vortex, gas in feed

Field Decision Tree

Step 1: Is NPSHa truly insufficient?

Calculate NPSHa at actual conditions, not nameplate:

NPSHa = (P_atm + P_gauge - P_vapor)/SG * 2.31 + Z_static - h_friction

Common errors:

Forgot to use vapor pressure at the actual pumping temperature - hot water at 200°F has Pv = 11.5 psia, not 0.5 psia at 80°F. This is the #1 mistake.
Friction head computed at nameplate flow, not actual flow - if the pump is running 30% above design flow, friction is roughly doubled (square law).
Strainer assumed clean - a fouled strainer can drop NPSHa by 5-10 ft. Check dP across the strainer in the field.

If NPSHa > NPSHr + 3 ft margin (per Hydraulic Institute ANSI/HI 9.6.1), classic cavitation is unlikely. Look elsewhere.

Step 2: Where is the operating point?

Pull the pump curve and overlay actual operating point. Most field cavitation problems happen because pumps are running 25-50% of BEP, not because NPSHa is short.

Operating below 50% of BEP - suction recirculation is likely
Operating below 40% of BEP on a high-suction-specific-speed pump (Nss > 11,000) - guaranteed recirculation
Operating past the right end of the curve - check for cavitation due to higher NPSHr at runout

Step 3: Is air getting in?

Check for these sources before tearing the pump apart:

Mechanical seal flush failure - flush is supposed to be product-side; if API Plan 11 has lost flow, atmospheric air can be drawn back
Loose suction flange gasket - may leak only under vacuum
Vortex at suction tank - check submergence per ANSI/HI 9.8
Two-phase flow upstream - flashing in a control valve, dissolved gas breakout in a low-pressure suction line

Step 4: Run a sanity vibration scan

If an accelerometer is on hand, do a 30-second velocity spectrum at the bearing housing:

Frequency Region	What You're Looking For
1x running speed	Imbalance, misalignment - not cavitation
Vane pass frequency (N x impeller blades)	Discharge recirculation if dominant; high pressure pulsation
Broadband above 5 kHz	Classic cavitation, mechanical contact, bearing wear
Sub-synchronous (0.5-1x)	Suction recirculation or surge

Fix Decision Matrix

Diagnosis	First Try	Permanent Fix
Insufficient NPSHa	Cool fluid 10-20°F, throttle minimum-flow line, raise tank level	Inducer, larger eye impeller, lower pump elevation, booster pump
Suction recirculation	Open up flow - bypass back to tank	Trim impeller down, smaller pump, two pumps in parallel each at BEP
Discharge recirculation	Open up flow - bypass back to suction (with cooling)	Trim impeller, replace with smaller pump
Air ingress	Rebuild seal flush, tighten gaskets, raise tank level	Mechanical seal upgrade (Plan 53A barrier fluid), self-venting suction

Heuristics That Save Time

Margin matters more than meeting NPSHr. NPSHr is defined as the value where head drops 3% (NPSH3) - that's already cavitating. For continuous duty, target NPSHa >= 1.3 x NPSHr and >= NPSHr + 3 ft, whichever is greater.
Hot water cavitates differently. Above ~180°F, vapor pressure rises faster than Reynolds-scaled inducer can handle. Use a colder API 11 flush or add an inducer rated for hot service.
Don't trust the sound from inside a pump house. Reverberation makes everything sound like cavitation. Walk to the suction piping and listen 6 ft upstream of the pump - cavitation noise is loudest at the impeller eye, not in the volute.
Damage location is your best evidence. Open the pump and look at where pitting is. Suction-side leading edge = NPSH cavitation. Pressure-side near discharge = discharge recirculation. Both = pump is way oversized for the actual duty.

References

ANSI/HI 9.6.1-2017 - Rotodynamic Pumps Guideline for NPSH Margin
ANSI/HI 9.6.3-2017 - Rotodynamic Pumps Guideline for Allowable Operating Region
ANSI/HI 9.8-2018 - Rotodynamic Pumps for Pump Intake Design
API 610 12th Ed (ISO 13709) - Centrifugal Pumps for Petroleum, Petrochemical and Natural Gas Industries
Karassik, I. Pump Handbook, 4th Ed - Ch 2.3 (Cavitation), Ch 8.3 (Vibration)