Fire Pump Testing & Alignment — Lessons from a Rebuild
Monthly runs are not just a check mark in the log book. Our fire pump passed every run for years — and still ended up needing a complete rebuild because nobody ever greased the bearings.

The pump that passed every test — until the panel said otherwise

On paper, this fire pump was a model citizen. Every month somebody walked into the pump room, pressed the start button, watched the electric motor spin up, let it run its ten minutes, and filled in the log. Suction pressure: same as last month. Discharge pressure: same as last month. Comments: “ran normal.” Twelve tidy rows a year, years running. If you audited the paperwork, you’d have called it one of the better-documented pumps in the portfolio.

Then the panel spoke up. The building engineer reported a trouble condition on the fire pump controller — the kind of signal that’s easy to write off as a nuisance and reset. Instead, it earned the pump a real visit. The technician who responded — from SPP Pumps, the pump’s manufacturer, whose Americas fire pump factory happens to sit in Norcross, Georgia — did what experienced pump people do before touching any paperwork: he ran the pump and listened. And the pump didn’t sound right. Nothing dramatic — just a note in the mechanical hum that told a trained ear something was dragging where it should be gliding. Years around these machines pointed him at exactly two suspects: the coupling alignment and the bearings. He rotated the shaft by hand, felt the coupling, checked the grease points. Then he delivered the verdict nobody wants on a Tuesday morning:

Here’s the uncomfortable part: every one of those monthly tests was “real.” The pump did start. It did run ten minutes. The gauges did read normal — churn pressure barely loads the bearings, so for years nothing screamed. It took a trouble signal on the controller and a technician who trusted his ears over the log book to surface what the paperwork never would. The failure wasn’t in the running of the test. It was in believing the test was the maintenance.

What the monthly (or weekly) run is actually for

Start with the frequencies, because they changed and half the industry still recites the old rule. Under current editions of NFPA 25, diesel-driven fire pumps run a no-flow (churn) test weekly for at least 30 minutes — the engine needs to reach operating temperature. Electric-driven pumps run monthly for at least 10 minutes NFPA 25 §8.3.1.2. A short list of electric configurations still requires weekly runs — most notably pumps protecting buildings beyond the pumping capacity of the fire department (high-rise territory) and certain vertical-turbine and limited-service-controller installations. Check which edition your state has adopted before you change any schedule.

But the frequency is the least interesting part. The run is an observation window — ten minutes when the machine is telling you everything about its health, if anyone is listening. During the churn test you are supposed to:

• 📊 Record the actual suction and discharge gauge readings — fresh numbers, read off the needles, not carried forward from last month’s sheet.
• 💧 Check the packing glands — a slight drip during operation is correct (it cools and lubricates the packing); a dry gland or a stream both earn a note.
• 👂 Listen and feel for unusual noise or vibration — a growling bearing or a buzzing coupling is audible months before it’s terminal. This is the check that would have caught our pump years earlier.
• 🌡️ Check the packing boxes, bearings, and pump casing for overheating — the back of a gloved hand near (not on) the bearing housings after the run tells you plenty.
• ♻️ Confirm the circulation relief valve discharges while churning — it’s the small valve that keeps a no-flow pump from overheating its own case.
• ⚡ On electric pumps — verify the controller starts the pump at its set pressure, and log the run time.

The maintenance table nobody reads

Buried after the testing sections everyone quotes, NFPA 25 §8.5 lays out the annual maintenance program — and it defers, repeatedly and deliberately, to the pump manufacturer’s instructions. The maintenance tables call for, among other items:

• 🛢️ Lubricate the pump bearings — annually as the floor, or on the manufacturer’s schedule if it’s more frequent (for many horizontal split-case pumps it is — commonly every six months or a set number of operating hours).
• 🔗 Lubricate the coupling — on coupling types that require it, per the coupling manufacturer.
• 📐 Check the pump/driver coupling alignment — annually. Alignment is a progressive condition: foundations settle, pipes pull, bearings wear. It is not an install-day checkbox.
• 🔩 Motor-side maintenance — motor bearing lubrication per the motor manufacturer, plus the electrical and controller items in the driver tables.
• 🧤 Packing adjustment — set for the correct drip rate, replaced when it can no longer be adjusted.

Two failure modes bracket the lubrication task, and both kill bearings. Never greasing is what happened to us — the grease dries and channels, metal contacts metal, the races spall, the shaft develops play, and eventually the pump binds. Over-greasing is the sneaky opposite: pumping a bearing full until the churning grease overheats it and blows the seals. The escape from both is boring and effective: open the manufacturer’s O&M manual, use the specified grease, the specified quantity, on the specified interval, and write down when you did it. NFPA 25 explicitly makes the manufacturer’s recommendation the governing document for maintenance method and frequency NFPA 25 §8.5.

From the technician’s field visit: the shaft end and bearing area up close. Where there should be clean, greased, protected metal, there’s corrosion and debris — the visible signature of a lubrication program that never happened.

Straight edge vs. the instrument: what each one can honestly tell you

When the coupling guard comes off, the classic move is the one in our hero photo: lay a machinist’s straight edge across the two coupling hubs and look for daylight, checking at four positions around the coupling, with a feeler gauge for the face gap. Let’s be fair to the old method: it is acceptable as a coarse check. It costs nothing, takes two minutes, and will catch gross misalignment — the kind you can measure in whole millimeters.

But be honest about the physics. Your eye judging light under a bar resolves maybe 10 thousandths of an inch on a good day. The straight edge sees only parallel offset, only in the plane you hold it. It cannot measure angularity — the two shafts meeting at an angle like a bent elbow — and angularity is the component that eats couplings and bearings fastest. It can’t separate vertical from horizontal error, and it produces no number you can write in a report or compare to last year.

The instrument on our pump: an SKF TKSA 11 shaft alignment tool chained to the shaft beside the coupling. This family of tools (laser-based systems work the same way; the TKSA 11 uses inductive proximity sensors) measures both shafts’ true centerlines as they rotate.

A precision alignment instrument — a laser system, or the inductive-sensor SKF TKSA 11 the technician used on our pump — works differently. Sensor heads mount on each shaft, the shafts are rotated together through three positions (roughly the 9, 12, and 3 o’clock sweep the app walks you through), and the tool computes the true relationship between the two shaft centerlines. Out comes the full picture in four numbers: vertical offset, vertical angularity, horizontal offset, horizontal angularity — resolved to a fraction of a thousandth of an inch, with live correction guidance for exactly how much to shim and how far to slide the driver.

The app walks the sweep: rotate the sensors to each clock position and record.

Probe distances read in thousandths (thou) — the unit alignment work lives in.

Our as-found numbers — and what “acceptable” actually is

The “As Found” screen on our pump. Four readings, four red X’s. The horizontal offset alone — 47.6 thou — is roughly twelve times the accepted limit for an 1,800 RPM machine.

Alignment work is measured in thousandths of an inch (“thou,” or mils), split into two kinds of error in two planes. Offset is the distance between the two shaft centerlines where they meet at the coupling — the shafts are parallel but not on the same line. Angularity is the angle between them, expressed as thou per inch of coupling diameter. Each exists both vertically and horizontally, which is why a complete measurement is four numbers.

NFPA 20 requires pump/driver alignment to the manufacturer’s specifications and Hydraulic Institute standards NFPA 20 §6.5; the widely used industry tolerance tables (ANSI/ASA S2.75 framework) set the limits by shaft speed. Here’s our pump against the 1,800 RPM column:

Two things are worth sitting with. First, a straight edge would not have quantified any of this — 17 thou of vertical offset is at the edge of what an eye can catch, and angularity is invisible to a bar entirely. Second, misalignment and bearing failure feed each other: bad bearings let the shaft wander, which worsens the effective misalignment, which loads the bearings harder. A commonly cited rule of thumb in the reliability world is that every additional mil of offset beyond tolerance on an 1,800 RPM machine takes roughly 10% off bearing life. Run the arithmetic on 47 thou and you understand why this pump’s bearings never stood a chance — grease or no grease.

How the realignment actually goes — and how long it takes

Brackets chained on both sides of the flexible coupling, straight edge still in place from the coarse check. The pump is aligned by moving the driver (the electric motor) — never the pump, which is fixed by its piping.

Because our pump ran bound and badly misaligned, the rebuild ends with a full precision alignment of the electric driver to the rebuilt pump. The sequence — whether it’s a laser system or the inductive tool in our photos — is the same:

• 1. Pre-alignment checks. Coupling condition, base and anchor bolts, piping strain, and a clean sweep under the motor feet. Garbage in, garbage out — a loose base bolt makes every measurement a lie.
• 2. Soft foot. Before any alignment move, verify each motor foot actually sits flat. Loosen each hold-down bolt one at a time and watch for movement (or slip a feeler gauge under the foot corners). A foot that springs is corrected with shims first — chasing alignment on a soft-footed machine is the classic way to spend three hours going in circles.
• 3. Mount and sweep. Sensor heads on each shaft, rotate together through the ~9-12-3 o’clock positions, and let the tool compute all four as-found values.
• 4. Vertical correction — shims. The tool tells you the exact shim change under each pair of motor feet. Pre-cut stainless shims only, and never stack more than three or four under a foot — a tall shim sandwich behaves like a spring.
• 5. Horizontal correction — the slide. With live values on the screen, the motor is nudged sideways with jacking bolts or careful persuasion until both horizontal numbers come in. This is where the live display earns its money — you watch the offset shrink in real time.
• 6. Re-measure, torque, re-verify, document. Final sweep with everything torqued (tightening moves things), then save the as-left report. That PDF is your annual baseline — next year’s check is a ten-minute comparison instead of an archaeology project.

Time budget, honestly: the measurement sweep itself takes 20–40 minutes once brackets are on. A routine annual verification on a healthy pump is about an hour including guard removal and reinstall. A corrective alignment typically runs 2–4 hours — most of it soft foot, shim work, and the iterate-measure-move loop. A post-rebuild alignment like ours sits at the top of that range or beyond — plan for up to half a day — because the driver is being repositioned from scratch, bolt holes end up bound, and old shim packs have to be rebuilt clean. Whoever budgets the rebuild should budget the alignment time with it; a rushed alignment puts the brand-new bearings right back on the old wear curve.

While the pump is down: don’t forget the impairment side

A fire pump rebuild is not an afternoon errand — parts, factory scheduling, and the realignment mean days out of service. That makes it a textbook system impairment: tag the pump controller, notify the AHJ, your insurance carrier, and the alarm monitoring company, brief the facility, and follow your impairment program (including fire watch where your policy or the AHJ requires it) until the pump is retested and returned to service. The rebuild fixes the machine; the impairment procedure is what protects the building while the machine is on the bench.

What you might not know — collected pump-room wisdom

Digging through pump-service blogs, reliability-engineering write-ups, and practitioner forums while researching this article turned up a handful of insights that surprised even people who’ve stood next to fire pumps for years. Consider this the “things the old-timers know” appendix:

• Pressing the RUN button doesn’t count as the test. NFPA 25 requires the no-flow test to be conducted by starting the pump automatically — dropping pressure in the controller’s sensing line (crack the sensing-line drain and watch the controller do its job) rather than punching the manual start. The run button only proves the motor spins; the automatic start on pressure drop is the exact event that has to work at 2 a.m. during a fire. A pump that’s been “tested” by button for years may never once have demonstrated the thing the test exists to demonstrate.
• Ten minutes at churn can cook a pump. At no-flow, all that horsepower has nowhere to go but into the water trapped in the casing, and the temperature climbs fast. The only thing standing between a churning pump and warped seals is the little circulation relief valve quietly dribbling hot water out so cool suction water replaces it. That dribble isn’t a leak — it’s the pump’s cooling system, and confirming it’s flowing is an explicit checklist item during every run. (Pumps with engine-cooled or special arrangements differ — know which yours is.)
• The wrong grease can be worse than no grease. Grease is a thickener holding oil, and thickener chemistries can be incompatible: mix a polyurea grease (common in electric-motor bearings) with a lithium-complex grease (what’s usually in the shop grease gun) and the blend can soften and lose its oil entirely — the bearing starves right after somebody “maintained” it. Reliability engineers also report that over-greasing kills more motors than under-greasing: excess grease churns, overheats, and gets forced past the seals into the motor windings. Match the grease to the manual, fill 30–50% of the cavity, and log what went in.
• “It was aligned at the factory” means almost nothing. Pump and driver ship on a common baseplate that’s aligned before it leaves — and baseplates flex. Rigging, transport, anchor-bolt torque, grout shrinking as it cures, and pipe flanges pulled into position all move the machines relative to each other. That’s why NFPA 20 makes alignment a field requirement after installation: the factory certificate is a quality-control record, not a description of your pump room.
• A bone-dry packing gland is a finding, not a win. The instinct to crank the gland down until the dripping stops destroys the seal: the drip (roughly a drop per second on a running pump) is what cools and lubricates the packing. Run it dry and the packing burns within days and scores the shaft sleeve — and a scored sleeve leaks past every new packing set you install afterward. Adjust a flat at a time, wait, re-check. Slow hands save sleeves.
• The diesel 30-minute rule is about chemistry, not thoroughness. A diesel that never reaches operating temperature doesn’t fully burn its fuel — the residue accumulates in the cylinders and exhaust as wet stacking, which degrades performance exactly when you need everything the engine has. Thirty minutes gets the engine to temperature, burns the deposits, turns over fuel, and lets cooling-system problems show themselves. Cutting the run short to save time quietly builds the failure you’re testing against.
• Your ears are a legitimate instrument. Bearings rarely fail without warning — they hum, then growl, then grind, usually over months. A mechanic’s stethoscope (or the old-school screwdriver-to-the-ear trick) pressed on a bearing housing isolates that progression long before it shows up on a gauge. It’s exactly how the technician in our story caught this pump: not with the log book, but with the sound of something dragging where it should be gliding. If the pump sounds different this month than last month, that is data — write it down.

The take-home checklist: monthly · quarterly · annual

Every run (monthly electric / weekly diesel)

• ▶️ Run the full required duration — 10 min electric, 30 min diesel NFPA 25 §8.3.1
• 📊 Record fresh suction/discharge readings — no carry-forward numbers
• 👂 Listen/feel: bearings, coupling, unusual vibration · 🌡️ check bearing/casing/packing-box temperature
• 💧 Packing gland drip correct · ♻️ circulation relief valve discharging at churn
• 📝 Log actual observations — a comment field that says something is worth ten that say “normal”

Quarterly (sprinkler-system ITM around the pump)

• 🔔 Test waterflow alarm devices · 🔥 inspect the FDC (caps, clapper, ID sign) NFPA 25 Ch 13
• 🚰 Main drain test where required quarterly (e.g., downstream of backflow/PRV arrangements)
• 🚪 Pump room housekeeping: heat working, ventilation clear, no storage creeping in around the pump

Annually — the ones that would have saved our pump

• 💦 Annual flow test at churn / 100% / 150% of rated capacity, plotted against the acceptance curve; more than 5% degradation demands investigation and correction NFPA 25 §8.3.3
• 🛢️ Lubricate pump bearings and coupling — manufacturer’s grease, quantity, and interval NFPA 25 §8.5
• 📐 Verify coupling alignment — with an instrument that produces numbers, compared to last year’s as-left report
• 🧤 Adjust/replace packing · 🔩 motor and controller maintenance per the driver tables
• 📁 File it all in the pump’s permanent record: flow curve, lubrication log, as-left alignment

For the full device-by-device frequency table across the whole sprinkler system, use our NFPA 25 ITM frequency tool — it’s the same data Clara cites in chat.

Inspection Report Language

If your churn tests are running clean but nobody can produce a lubrication or alignment record, that absence is the finding. Copy/paste starting language:

Ask Clara

Not sure whether your pump logs would survive this article? Clara — the site’s assistant — can walk your specific pump (electric or diesel, your state’s adopted edition) through the required frequencies, the churn-test observation list, and what a defensible maintenance record looks like.