How to Record a Pump Installation So the Next Visit Is Shorter

Why this matters

Most pump records are identification: make, model, serial, impeller size, date. All of that is recoverable later off a nameplate or a curve sheet, so none of it shortens a future visit. What is not recoverable is what the machine did on the day it was new and un-fouled, and six readings on that list exist only in that window. Every later diagnosis on that pump is a subtraction against one of the six. Miss them and the next tech has to reconstruct a baseline from nothing, which is what turns a forty-five minute call into most of an afternoon.

This article is the record itself, field by field, with the reason each field exists and what a later visit does with it. The last section is one filled in and then used eighteen months later.

Safety before any reading is taken

Take pressure readings from installed gauge taps only. Do not crack a fitting, a plug or a union on a pressurized line to get a reading; pumped liquid escaping a partly loosened joint injects and, on a hot-water or condensate line, flashes and scalds. If there is no tap, the record entry is "no tap fitted" and the corrective action is to fit one during a planned shutdown.

Anything that puts a hand, a tape flag or a dial indicator on the shaft happens with the machine locked out. Open the motor disconnect, apply lock and tag, try the start command, and then relieve residual energy in the casing by closing the suction and discharge valves and opening the casing vent, which is the stored-energy step 29 CFR 1910.147(d)(5)(i) requires after the isolation devices are locked. Locking the disconnect to stop rotation is a mechanical isolation under 1910.147; if the work extends to opening the starter enclosure and metering conductors, that act is excluded from 1910.147 at (a)(1)(ii)(C) and is governed by 29 CFR 1910.333(b)(2) in general industry or 29 CFR 1926.417 in construction.

The six perishable fields

These are the ones that cannot be recovered later. Everything else in the record can wait.

1. Duty point as installed. Suction gauge and discharge gauge, read together, with the position of every throttling valve in the loop written down beside them. A pressure pair with no valve position attached is not a duty point, it is two numbers. Later use: a discharge pressure that has fallen with the same valves in the same position is a pump or an impeller problem; one that has fallen with a valve moved is a system problem, and you will not be able to tell which without this field.

2. Motor current at that duty point, with nameplate full-load amps written beside it. One clamp reading on one lead, and note which lead. Later use: on a centrifugal pump current tracks flow, so a drop in current against this baseline says the pump is moving less liquid, which is the fastest confirmation there is that a suction-side complaint is real.

3. As-new differential across the suction strainer or filter, at a stated flow. This is the field most often skipped and the one with the sharpest payoff, because a strainer has a pressure drop when it is perfectly clean and that drop is not fouling. Later use: the fouling contribution is the measured differential minus this baseline, corrected to the flow at which the later reading was taken. Both corrections matter and the worked example below shows one of them flipping a decision.

4. Cold alignment values and the thermal growth allowance used. The dial or laser readings you left the machine at, plus the deliberate cold offset you set so it comes into alignment hot. Later use: a machine that reads misaligned cold may be correctly aligned, and without this field the next tech will "correct" the offset and create a hot misalignment. The sibling article on allowing for thermal growth when you align cold owns the method; this field just preserves the number it produced.

5. Shutoff head, where the manufacturer's own documentation permits it. With a centrifugal pump only, and only if the pump documentation allows a brief closed-discharge reading, close the discharge for a few seconds, read the discharge gauge, and reopen. Some manufacturers forbid it outright and that document governs. Never do this on a positive-displacement pump, where blocked discharge has nowhere to send the flow, and never on a hot or flammable liquid, where the trapped volume heats with every second the valve is shut. Later use: shutoff head is a fixed point on the curve that identifies the impeller you actually have, which is how you catch a replacement impeller that was trimmed differently.

6. What it sounds like when it is right. One line of plain description plus the coast-down behaviour: whether any noise stops the instant flow stops or decays with rotation. Later use: this is the comparator the analyzer-free vibration method needs, and the sibling article on reading a pump vibration complaint spends its first step manufacturing it because most records do not carry it.

The fields that are worth writing but are not urgent

Nameplate data, curve sheet with the duty point marked on it, impeller diameter as supplied, coupling type and insert class, seal arrangement and flush plan, suction piping description in straight-run pipe diameters ahead of the suction flange, and the isolation and bleed points a future lockout will use. All recoverable, all worth five minutes, none of them time-critical, because a nameplate will still be there in three years.

The suction piping description deserves a sentence of its own. Write the number of pipe diameters of straight run ahead of the suction flange and what is immediately upstream of that, because a pump fed by an elbow directly on its suction flange has a permanent, non-degrading flow-distortion problem that will present, years later, as intermittent noise and short seal life. If that entry says "elbow on the flange" the next tech stops looking for a fault that developed.

What the record is not

It is not a warranty claim file and it is not an inspection report. It carries no judgement, no pass or fail, and no recommendation. Fields it should deliberately exclude:

• Anything requiring a tool the shop does not own. A record with an empty vibration-velocity column teaches the next reader that the record is unreliable, which is worse than not having the column.
• Derived values. Write the two gauge readings, not the calculated head. A later reader can compute head from the readings; nobody can recover the readings from a head figure, and a transcription error in a derivation is silent.
• Anything already on the curve sheet. Attaching the curve is one action. Copying six points off it by hand is six chances to be wrong.

The record, filled in

A 120 gpm end-suction centrifugal, new install, lead of a lead-lag pair.

Time to capture: about 0.4 on-site labor hours, most of it waiting for the gauges to settle.

Eighteen months later, the record does the work

Complaint: the lead pump is noisy when the lag pump is off. On arrival the lag pump is off, so this pump is running further out on its curve than the commissioning point, and the strainer differential reads 4.6 psi at 160 gpm.

The general rule this record sets: clean the element when the fouling contribution exceeds 3.0 psi above the corrected as-new baseline, measured per element at a stated flow. That 3.0 psi is a shop default, not a manufacturer's limit; tune it to your own history and to any figure the strainer's own documentation gives.

Two corrections, both stated before the reading is interpreted, both printed:

• Raw differential: 4.6 psi at 160 gpm.
• Correction one, the as-new baseline: 1.3 psi at 120 gpm. It has to be subtracted, because a clean element is not a zero-drop element.
• Correction two, the flow basis: the baseline was taken at 120 gpm and the reading at 160 gpm, so they are not comparable as printed. A coarse wire-mesh strainer behaves like an orifice, so its clean drop rises roughly with the square of flow: 1.3 psi times (160/120) squared is 1.3 times 1.78, which is 2.3 psi. That square law is derived for an orifice-like mesh in turbulent flow; a fine filter element is a porous medium whose clean drop is closer to proportional with flow and with viscosity, so an element gets a linear correction, not this one. This machine has a mesh strainer, so the square law governs.
• Fouling contribution: 4.6 psi minus 2.3 psi is 2.3 psi.
• Decision: 2.3 psi is below the 3.0 psi rule. The strainer is not the complaint.

Look at what each correction was worth. Raw against the rule, 4.6 psi against 3.0, says clean it now. Baseline subtracted but flow uncorrected, 4.6 minus 1.3 is 3.3 psi against 3.0, still says clean it now. Only with both corrections applied does the answer change, and it changes to the correct one. A tech who cleaned the strainer would have found it lightly loaded, declared the job done, and been back inside a month, because the noise is a suction-side condition created by running out on the curve with the lag pump off - which the duty-point field and the sound field together identify in the next ten minutes.

What the record was worth in hours. That visit closed in about 0.75 on-site labor hours. The same complaint on a pump with no baseline runs about 2.5 on-site labor hours in this shop's own history, most of it spent establishing what normal was, and it ends in a strainer clean roughly half the time. Against the 0.4 hours the record cost to capture, the first call it shortens returns more than four times the time it took, in the same unit of on-site labor hours, and the record keeps paying on every visit after that.

The failure mode. The most common way this record fails is not omission, it is a field written without its condition: a strainer differential with no flow, a gauge pair with no valve position, an alignment offset with no note that it is a deliberate cold offset. Each of those looks complete on the page and is unusable, and the next tech will use it anyway, because a number written down is trusted more than a blank.

References

• 29 CFR 1910.147, the control of hazardous energy, including the relief of stored and residual energy at (d)(5)(i), for any commissioning step that puts hands on the shaft or opens the casing
• 29 CFR 1910.147(a)(1)(ii)(C), the exclusion that sends work on energized conductors to 29 CFR 1910.333(b)(2) in general industry and 29 CFR 1926.417 in construction
• Pump manufacturer documentation for the performance curve, the permitted shutoff-head reading and any strainer or element differential limit, which own those numbers
• See related: How to Read a Pump Vibration Complaint Without an Analyzer; How to Allow for Thermal Growth When You Align Cold; What a Pump Curve and a System Curve Do Together