How Suction Conditions Fail in Practice

Why this matters

Almost nobody installs a pump with bad suction conditions. Pumps get installed with margin, run fine, and then lose that margin one term at a time over a couple of years while nothing visible changes. By the time the noise starts, the tech is looking at an install that was correct on day one, so the pump gets blamed and replaced, and the replacement does the same thing. The useful skill is not calculating suction head from scratch. It is knowing that the number is a running budget of four terms and being able to say which term moved.

Before you open anything on the suction side

A suction line can be under vacuum, and the action is to break vacuum deliberately at a high-point vent before you loosen a flange, not by cracking the flange itself. A joint opened under vacuum pulls air and debris inward and can suck a gasket into the line; worse, on a hot line, the moment atmospheric pressure enters, liquid that was stable under vacuum flashes to steam.

Hot liquid above scald temperature is the other one, and the action is to isolate, verify the isolation holds by watching a gauge for a minute, and let the section cool before breaking a joint. If schedule will not allow cooling, drain to a routed hose rather than to the floor, stand out of the plane of the joint, and use a face shield rather than glasses.

If you are opening the pump itself rather than the piping, that is mechanical isolation and stored energy under 29 CFR 1910.147: lock and tag the disconnecting means rather than switching it off, and relieve trapped pressure before the first joint. Where the fluid is a glycol mix, a treated boiler water or a process fluid, read its SDS before you open a hot line, because the hazard route from a flashing line is inhalation of vapor or aerosol rather than skin contact; if the SDS calls for respiratory protection, that puts you inside the employer's respirator program at 29 CFR 1910.134, which requires fit testing before first use and at least annually thereafter under 1910.134(f)(2). A glove is the wrong control for a vapor.

The budget has four terms and only four

Available suction head, expressed in feet of the fluid you are actually pumping, is:

absolute pressure on the liquid surface, plus static head, minus suction friction, minus the fluid's vapor pressure at its actual temperature.

Static head is positive when the source is above the pump centerline (flooded suction) and negative when the pump has to lift. Everything is in feet of that fluid, which means the conversion from psi depends on specific gravity: cold water is about 2.31 ft per psi, and water near 180 F, at a specific gravity of roughly 0.97, is about 2.38 ft per psi.

Required suction head comes off the pump curve, at the flow you are actually running, and it is not a pass or fail line. The published value is defined as the point at which the pump's developed head has already dropped by 3 percent, which means the pump is already cavitating there. Running at exactly the required value is running at the documented onset, not at a safe point. Common specification practice is a margin of at least a few feet, and many jobs use a ratio of 1.2 to 1.5 times the required value; the pump manufacturer's own guidance governs where it differs.

Which term actually moves after commissioning

• Atmospheric term. Fixed at a site, but not the same at every site. Standard atmosphere at about 5,000 ft elevation is roughly 12.2 psia against 14.7 at sea level, which for hot water is around 6 ft of head gone before the pump does anything. This term also moves on a closed loop, where the surface pressure is the fill pressure rather than atmospheric, so a leaking expansion tank or a fill valve set low moves it directly.
• Static head. Moves whenever tank level moves. A design based on the tank's normal level rather than its low-level cutout is a budget that fails only at the end of a draw-down, which is exactly when nobody is watching.
• Suction friction. The term that moves most often and grows with roughly the square of flow. A strainer basket, a partly closed suction isolation valve, a collapsed flex connector, a foot valve with debris on the seat.
• Vapor pressure. Moves with fluid temperature, and moves faster than people expect. This is the term that makes hot systems, condensate, boiler feed and heat-recovery loops the hard cases.

Worked budget: the same pump, two years apart

An open-tank transfer pump at sea level, water at 180 F. Specific gravity about 0.97, so about 2.38 ft per psi. Tank low-level is 4 ft above the pump centerline. Vapor pressure of water at 180 F is roughly 7.5 psia. The pump's published required suction head at design flow is 12.0 ft.

At commissioning, 18.6 available against 12.0 required is a ratio of 1.55, inside normal practice. Two years later it is 6.2 against 12.0, a ratio of 0.52, and the pump is not marginal, it is well past the documented onset.

Two things changed, and the one that mattered was not the one that was noticed. A strainer that was never cleaned took suction friction from 2.5 ft to 8.0 ft, costing 5.5 ft. A process change took the tank from 180 F to 195 F, where vapor pressure is roughly 10.4 psia, or about 24.8 ft, costing 6.9 ft. The temperature change is 56 percent of the 12.4 ft that disappeared and the strainer is the other 44 percent, and 18.6 minus 12.4 is the 6.2 ft in the table.

The failure mode here is real and common: the tech cleans the strainer, recovers 5.5 ft, gets to 11.7 ft against 12.0 required, and the noise mostly goes away at part load. The job is signed off. The pump is now running with a ratio of 0.98 at design flow, still under the published required value, and it will chew its impeller quietly for another year. Recovering one term is not the same as restoring the budget.

What would have flipped the recommendation: if the temperature rise were permanent and the strainer were already clean, no amount of suction-side cleanup gets 12 ft back, and the honest answer is a pump with a lower required suction head, a raised source, or a suction line one size up, not another service call.

Why "it worked for two years" is not evidence

Every term in that budget is measured at a condition, and the condition that matters is the worst one the system actually reaches, not the one you can see on the day. Three conditions routinely produce a system that is fine on inspection and cavitating in service:

• Peak flow. Friction rises with roughly the square of flow, so a suction line that costs 2 ft at half flow costs around 8 ft at full flow. A morning walkthrough at low demand reads clean.
• Low tank level. The static term is at its worst at the low-level cutout, which happens for a few minutes at the end of a draw-down.
• Highest fluid temperature. On any system with heat recovery, seasonal setpoints, or a process that has crept, the vapor-pressure term is worst on the hottest day at the highest setpoint.

If all three land in the same ten minutes, and on a lot of systems they do, that is the condition your budget has to survive.

Where the method changes

Closed hydronic loops. The surface-pressure term is fill pressure at the expansion tank, not atmosphere, and the tank's location relative to the pump decides where the pump adds or subtracts pressure around the loop. A waterlogged expansion tank and a circulator pumping toward the point of no pressure change produce inlet cavitation on a system that looks correctly filled at the gauge.

Positive-displacement pumps. They still have a required inlet condition, and it is usually stated as required inlet pressure rather than head, but the sensitivity is different: they are more tolerant of low absolute pressure and far less tolerant of viscous fluids and long, small suction lines, because the inlet has to fill a fixed volume in a fixed time. Slow it down and the requirement drops.

Self-priming and submersible pumps. A self-priming pump solves the priming problem, not the suction-head problem; once running, the same four terms apply. A submersible sits in the fluid, so the static term is large and positive and the usual failure is temperature or a blocked intake screen instead.

How to verify you got this right

Fit a compound gauge at the pump inlet, as close to the flange as the tapping allows, and take all four readings under the worst combination the system will actually see: peak flow, low tank level, highest setpoint temperature. A reading at any easier condition tells you nothing about the condition that fails.

Convert the gauge reading to absolute by adding local atmospheric pressure, subtract the vapor pressure of the fluid at the temperature you measured at the same time and place, and convert to feet using the specific gravity at that temperature rather than at 60 F. Compare against the pump's published requirement at the flow you were actually at, not at design flow, because the requirement rises with flow.

Then write both numbers and the conditions on the equipment record. A suction budget with no stated flow, level and temperature is not a measurement, it is a note, and the next tech will have to start over.

References

• 29 CFR 1910.147, control of hazardous energy, for lock and tag plus pressure relief before opening a pump
• 29 CFR 1910.134, respiratory protection, including fit testing before first use and at least annually under 1910.134(f)(2), where an SDS calls for respiratory protection on a hot or volatile fluid
• Hydraulic Institute standards for net positive suction head terminology, including the 3 percent head-drop definition of the required value
• Published steam tables, or the fluid supplier's property data, for vapor pressure and specific gravity at the actual operating temperature
• See related: What Cavitation Is and How It Announces Itself; Reading Pump Curves Reference; Why Throttling a Valve Changes More Than Flow