The 9-stage Flowserve MSN-type pump shaft fractured in the threaded region at the throttle-sleeve corner; that location makes the thread root and its operating stresses the first investigation targets, but it does not identify a root cause by itself. The pump ran at 3,560 rpm, 190 °F, 30 psig suction pressure, and 1,500 psig discharge pressure, with reported flow within 15% of BEP and no vibration issue before failure. Use the checks below to separate a thread-root fatigue process from a transient overload, reverse rotation, or installation stress before selecting a repair.
Check the fracture origin before assigning a cause
- Prerequisite: Preserve the shaft and sleeve in their as-found condition. Record orientation, fracture location, thread engagement, and sleeve position before cleaning or machining. Photograph the fracture and adjacent surfaces at useful magnification. Gate: If the fracture origin is visible at a thread root or sleeve corner, document that exact location and proceed to step 2. If it is obscured, arrange a controlled failure examination before removing material.
- Inspect both fracture faces and nearby thread flanks for a progressive crack region, rubbing, fretting, corrosion, and a final overload region. Treat the reported fretting/staining and clean final fracture appearance as an observation to verify on the retained parts, not as a confirmed fracture diagnosis. Gate: A confirmed progressive region supports fatigue; a predominantly sudden fracture supports overload. Either result still requires the operating-condition checks below.
- Compare the fracture face and thread-root condition with the penetrant-test record. The reported liquid-penetrant test found no other crack. That result does not rule out an internal flaw or establish that the fracture developed suddenly; penetrant testing detects surface-breaking indications in the inspected condition. Gate: If the fracture origin remains uncertain, consult Flowserve or a qualified failure-analysis laboratory for examination before repair disposition.
Compare recorded duty with the pump operating point
- Collect the operating record nearest the failure: flow, suction and discharge pressure, temperature, speed, vibration trend, and any control or trip event. Confirm that pressure readings and flow correspond to the failed pump and the same operating interval. The reported values are 30 psig suction, 1,500 psig discharge, 190 °F, and 3,560 rpm; reported flow was within 15% of BEP.
- Plot the actual flow and head against the pump curve and design duty point. Use the applicable curve and the correct fluid properties to evaluate head; pressure alone does not define pump head. Gate: If the recorded point is outside the curve or away from the intended duty, investigate the operating condition and its duration before concluding that the shaft or sleeve alone caused the failure. If the point remains close to the specified duty, continue to the transient check.
- Trend vibration rather than relying only on an operator's recollection that no vibration issue existed. Check for changes in vibration, bearing condition, rub evidence, and any rotor-to-stationary-part contact. Reported wear-ring running clearance was 0.017–0.020 in.; compare the measured values with the pump's applicable limits and the as-found rub pattern. Gate: Rub marks or a changed vibration trend direct the investigation toward rotor position, alignment, and hydraulic/mechanical loading; no such evidence makes a transient or localized thread-root stress more important to test.
Test for a switching or reverse-flow transient
- Prerequisite: identify the failed pump's start/stop sequence, the sister pump's status, discharge check-valve condition, and available pressure or speed traces. The pumps were alternated to confirm operability, but the sequence and valve response at the failure are not specified.
- Review event logs and operator records for a rapid pressure excursion, abrupt check-valve closure, failed or delayed closure, or a start while the impeller was rotating backward. Water hammer can impose a short-duration load that steady-state pressure readings will not reveal. Reverse flow through a stopped pump can drive reverse rotation; restarting against that condition can impose a transient torque on the rotor and shaft.
- Inspect the relevant check valve and discharge piping for sticking, blockage, leakage, or evidence of delayed closure. Gate: A confirmed pressure spike or reverse rotation advances the transient as a credible contributing cause; correlate it with fracture morphology and event timing. If records show no transient, do not rule one out solely from normal steady readings—determine whether the installed instrumentation captures the switching interval, then continue with the alignment and sleeve checks.
Separate alignment and pipe strain from hydraulic loading
- Check the alignment record from the repair performed about a year before failure and compare it with current cold and operating alignment measurements, as applicable to the installation. Inspect soft foot, base condition, coupling condition, and piping strain. A long multistage casing can be sensitive to misalignment or distortion, and those conditions need not produce an obvious vibration complaint.
- Measure alignment using the applicable pump and driver procedure, then check whether connected piping forces move the equipment from its unstrained position. Record the actual readings and the specified acceptance limits from the equipment documentation. Gate: If readings change when piping strain or soft foot is corrected, investigate that displacement as a recurring shaft-bending load; if alignment remains within the applicable limits under the measured condition, proceed to the sleeve and thread assembly.
- Inspect wear rings, center-stage components, and other internal clearances for rub marks or uneven contact. The reported 0.017–0.020 in. clearance is not, by itself, proof of correct rotor position or alignment. Gate: Circumferential or localized rub evidence supports a rotor-position investigation; absent rub evidence does not exclude external bending or a short transient.
Verify the sleeve joint and prior repair details
- Establish how the throttle sleeve is attached: verify whether it is threaded, interference-fit, tapered, or another arrangement from the assembly drawing and the actual parts. The shaft fractured in a threaded area, but the fastening method and tightening procedure were not reported. Do not infer the joint design from appearance alone.
- For a threaded joint, inspect the shaft and sleeve threads at the root and first engaged threads for damage, fretting, galling, uneven contact, and evidence of incorrect engagement. Compare thread form and dimensions with the drawing. Check the specified tightening method and assembly records; use the manufacturer’s procedure rather than an assumed torque. Gate: Damaged or uneven thread contact, incorrect assembly, or a crack origin at a thread root directs corrective work to the shaft/sleeve joint and its stress concentration. If the joint is not threaded, follow the documented fit and retention procedure for the actual design.
- Review repair records for shaft reuse, sleeve replacement, machining, material traceability, and any changed fit or alignment. The shaft was not replaced at the last repair, approximately one year before the failure. That timing makes the repair history relevant, but it does not prove the repair caused the break. Have Flowserve review the shaft, sleeve, failure location, and operating record when the design-specific limits or assembly method are unclear.
Use the evidence to rank competing explanations
| Observed clue | What it supports | Next discriminating check |
|---|---|---|
| Fracture at the threaded throttle-sleeve region | A local stress concentration is a plausible crack-initiation location; the fracture location alone does not prove fatigue. | Map the crack origin and progressive fracture region; examine thread condition and assembly. |
| Reported fretting/staining with a clean fracture area | Could indicate progressive damage followed by final overload if confirmed on the parts. | Verify the morphology through controlled fracture examination. |
| Near-BEP reported flow and no reported vibration issue | Does not establish freedom from short transients, pipe strain, or localized bending. | Compare pump curve and operating logs; inspect alignment, casing internals, and transient records. |
| Wear-ring clearance of 0.017–0.020 in. | Provides an as-reported clearance measurement, not a stand-alone finding of acceptable rotor position. | Compare with the applicable design limit and check for rub marks. |
| No other crack found by liquid penetrant | No other surface-breaking indication was reported in the inspected condition. | Use fracture examination to assess the origin; do not treat the test as proof against internal flaws. |
Fatigue at a thread root followed by final overload is a working explanation only if the fracture examination confirms progressive crack growth. A transient or recurring bending load can supply cyclic stress; the thread root can concentrate that stress. A sudden event can instead produce overload, and the possibilities can coexist: a fatigue-weakened section may fail during a transient. Rank causes by matching the fracture evidence to records and measurements, not by selecting one from the break location alone.
Close the resolving branch with a controlled repair
- Set the disposition from the confirmed branch: correct a documented valve or switching problem; correct alignment, soft foot, or pipe strain when measured; or repair the sleeve/shaft joint when its fit, thread condition, or assembly is defective. If the fracture origin or material condition remains unresolved, obtain a design-specific review before returning the pump to service.
- Replace or rework parts only to the applicable Flowserve drawing and documented repair procedure. Record shaft and sleeve dimensions, fit or thread condition, rotor clearances, alignment readings, and the approved assembly method. Do not reuse a shaft with an unresolved crack origin or damaged thread root.
- Before startup, verify free rotation, completed assembly checks, alignment, connected piping condition, valve function, and instrument availability. During a controlled startup, capture flow, suction/discharge pressure, speed, temperature, vibration, and the switching transient. Compare readings to the pump curve and equipment limits; stop and investigate any unexpected rub, pressure excursion, reverse rotation, or vibration change.
Frequently asked questions
How do I determine whether a pump shaft fracture started at a thread root?
Preserve and examine both fracture faces, map the origin, and inspect the engaged threads for damage and fretting. Confirm a progressive fatigue region through controlled failure examination rather than relying on the fracture location alone.
How do I check whether a pump switch caused shaft overload?
Review the switching sequence, pressure and speed traces, and discharge check-valve condition for abrupt closure, a pressure spike, or reverse rotation before restart. Steady readings of 30 psig suction and 1,500 psig discharge do not show what happened during the switching interval.
How do I interpret 0.017–0.020 in. pump wear-ring clearance?
Compare the measured clearance with the applicable pump documentation and inspect for rub marks. The measurement alone does not establish acceptable rotor alignment or rule out pipe strain and transient loading.
How do I verify the repair before returning the pump to service?
Record the shaft/sleeve assembly condition, alignment, piping condition, clearances, and valve function against the approved procedure, then capture operating readings during controlled startup. Confirm flow against the pump curve and check that pressure, speed, temperature, and vibration remain within applicable equipment limits.