Why is it costly to open a failed pump or compressor right away and fit a spare?
Opening a failed pump or compressor right away, finding the broken part and fitting a spare is correct for production but costly for evidence, because much of what a pump or compressor records about its own failure resides in no single component. That record resides in relationships between parts: where the rotor sat axially, how much float remained, which clearance closed first, which direction a rub ran, and what was lying in the casing.
The relationships between the parts of a failed pump or compressor exist only while the machine is assembled, and teardown, cleaning and rebuild erase them in that order.
Why does the position of parts matter when working out how a pump or compressor failed?
The position of parts matters because a pump or compressor failure is a sequence, and the component that broke most conspicuously is frequently not the one that failed first. Establishing the order of a pump or compressor failure requires reading the machine as an assembly: the rotor’s axial and radial position as found, the phase relationship between rub marks on rotating and stationary parts, and which labyrinth closed and in what direction.
None of that positional evidence in a pump or compressor survives a rotor pull unless it is captured first. Match-marking halves, indexing the rotor before anything is loosened, and photographing each interface before it separates costs an hour and preserves the failure sequence.
Can shaft alignment and pipe strain on a failed pump or compressor be measured after it is uncoupled?
No. Shaft alignment on a failed pump or compressor can only be measured with the machine as found and the coupling intact; once the coupling is split, any later alignment figure describes a new condition, not the one that failed. Misalignment is among the most commonly alleged contributors in pump and compressor failure matters and the most easily rendered unprovable.
Pipe strain on a failed pump or compressor is the same. Loosening the suction and discharge flanges and measuring how far the machine moves is a one-time test that separates a piping problem from a machine problem. Soft foot, the shim stack and the state of the grout are equally perishable. American Petroleum Institute (API) RP 686 sets out the installation practices these as-found readings are compared against.
What evidence does cleaning destroy in a failed pump or compressor?
Cleaning a failed pump or compressor destroys deposits, coke, polymer and product residue, which in a failure investigation are not dirt but evidence of what the machine handled and how hot it became. Heat tint records the temperature a rub reached. Solvent-washing a seal face removes the film identifying the fluid the seal saw, and bead-blasting an impeller destroys pit morphology outright.
Strainer contents, the debris field in the casing and the oil in the reservoir of a failed pump or compressor are samples rather than waste. Oil analysis for babbitt, water and wear metals is often the only surviving record of when a bearing began shedding material.
What can a failed mechanical seal reveal about a pump or compressor failure?
A failed mechanical seal from a pump or compressor can reveal loss of the fluid film, indicated by heat checking and blistering, and misalignment or distortion, indicated by an uneven wear band around the face circumference, while coking, chipped outside diameters and elastomer swelling each narrow the field. A mechanical seal carries a legible history. Mechanical seals are consumables and evidence at once. American Petroleum Institute (API) 682 defines the seal arrangements and support systems these observations are read against.
The trap with mechanical seals is commercial rather than technical. Failed cartridge seals routinely go back to the supplier on a core exchange for credit, removing the most informative component within days, usually before anyone has decided there is a case.
What does the thrust bearing show about how a pump or compressor failed?
The thrust bearing records the direction the rotor was pushed: which thrust pads are damaged, active face or inactive, shows that direction, and a thrust excursion is the signature of a transient rather than gradual wear. Within a pump or compressor’s failure chain, the thrust bearing is the most articulate part.
Reading a thrust bearing depends entirely on knowing which pad came from which position. A bag of loose thrust pads is a fraction as useful as the same pads tagged by position and orientation.
Can the internal clearances of a failed pump or compressor be measured after the rotor is pulled and parts are machined or scrapped?
No. Once the rotor of a failed pump or compressor is pulled and components are machined or scrapped, the wear-ring, throat bush, interstage and labyrinth clearances, balance-drum condition and rotor float cannot be recovered from anything. All of those are as-found values that can be measured once. Recorded against OEM tolerance, the clearances of a failed pump or compressor show how far the machine had drifted; unrecorded, they become a matter of recollection.
The same applies to the rest of the machine train. Gearbox backlash and contact pattern, coupling condition and the state of any spacer belong to the story of a pump failure even when the failure is inside the pump.
Does the operating data from a failed pump or compressor survive long enough to investigate the failure?
Often not. The operating data from a failed pump or compressor expires sooner than the hardware: vendor archives, alarm logs and operator shift records have their own retention windows, most shorter than the interval between a failure and the decision to investigate it. Process historians compress and eventually roll off. Machinery protection systems built to American Petroleum Institute (API) 670 hold transient and startup captures in buffers of finite depth, and those buffers overwrite on the next event.
A written preservation request covering historian tags at native resolution, vibration waveform and orbit data, event logs and maintenance records is worth issuing before a failed pump or compressor is opened.
How should a failed pump or compressor be taken apart when more than one party may be responsible?
Where more than one party may be responsible — an OEM, a repair shop, an installer, an operator, an insurer — the teardown is best conducted under an agreed protocol: notice to interested parties, an agreed sequence and photography, no destructive step without consent, and chain of custody for removed components.
The repair shop is frequently one of the parties that may be responsible. A repair shop’s strip report is useful but is no substitute for independent observation, because the repair shop is describing work it may later have to defend.
Can evidence be preserved when a failed pump or compressor has to go back into service quickly?
Yes. Production pressure is real and rarely negotiable, but the evidence steps that matter most when a failed pump or compressor has to go back in anyway are cheap and fit inside a single shift: photograph as found, measure alignment and pipe strain before uncoupling, bag and tag by position, take oil and debris samples, and freeze the data.
Returning the plant to service and preserving the answer to why a pump or compressor failed are not in genuine conflict. They only appear to be when preservation is attempted afterward, by which point most of the evidence is no longer available at any price.
This article is general technical orientation, not a failure analysis, an engineering opinion, or advice on any specific matter. Determining the cause of a particular incident requires hands-on examination by a credentialed expert.