A Standby Pump Can Show "Available" on Every Checklist and Still Not Start When You Need It

A standby pump, blower, or other backup unit at most ETPs and STPs gets confirmed "available" the same way, day after day: it is physically present, powered, connected to the process, and listed on the equipment roster as the duty unit's backup. That check is real and it catches genuine problems, a missing unit, a disconnected cable, a breaker left open. It does not confirm the condition that actually determines whether the standby unit will do its job when called on, because a pump can be present, powered, and connected while still being mechanically or electrically incapable of starting and running under real load, and a presence check has no way to notice the difference.

Being present and being ready are different claims

A presence check answers: is the standby unit physically in place and electrically connected. An exercise test, actually starting the standby unit and running it under load, answers a completely different question: will this unit actually do the job if called on right now. A pump can pass the first check indefinitely while failing the second, because the failure modes that prevent a pump from starting, a seized impeller from prolonged inactivity, a failed starter contactor, a dead control battery, debris lodged in the volute, corrosion on electrical contacts that never degrades function while sitting idle but prevents a clean start, are invisible to a walkthrough that only confirms the unit is there and powered. Redundancy built into a system is only real redundancy if the backup component actually works when called on, and a standby unit that has silently failed while reporting as available provides none of the reliability its presence implies.

Why this exact failure mode is built to stay hidden until the worst moment

A standby unit that never runs produces no symptom that a presence check, or even most routine inspections, would catch. It is not generating abnormal readings, because it is not running at all. It is not showing wear, because it is not being used. The only way its failure becomes visible under ordinary operating conditions is if the duty unit fails and the standby is called on to take over, at which point a standby that has quietly failed is discovered failed at precisely the moment its function was needed most, with the primary unit already down and no working backup behind it. This is the textbook description of a dormant or latent failure in reliability engineering: a fault in a standby or backup path that produces no symptom during normal operation and is only discovered on demand, exactly when discovering it is most costly. A standby unit shares this blind spot with standby power generators, which the same logic applies to directly: a generator correctly sized for its starting load is still only a real backup if it is periodically run under meaningful load rather than left to sit, since wet-stacking and similar idle-related degradation are well documented consequences of standby engines that are never actually loaded.

This is the same shape of mistake we described in the piece on analyzer sample transport lag: a check that is correct and genuinely useful against the narrow question it was built to answer can leave a completely different, more consequential question unexamined, because nothing in the routine ever asks the second question. There, calibration confirmed accuracy while sample timeliness went unmeasured. Here, a presence check confirms the standby unit exists and is connected while its actual ability to start and run under load, the only thing that matters in an actual failover, goes untested for as long as nobody deliberately runs it.

Picture a plant with a duty and standby pump pair at a critical lift station, the standby unit present, powered, and logged as available on every shift walkthrough for months. The standby has not actually been started in that time, because nothing in the routine calls for starting a unit that is not needed while the duty pump is running fine. During that period, the standby's impeller develops light seizure from prolonged inactivity in a wet environment, a failure mode invisible from outside the casing and undetectable by any check that does not actually attempt to run the unit. When the duty pump eventually fails, the standby is called on and does not start, and the plant discovers, at the exact moment it most needed functioning redundancy, that its redundancy had been nonexistent for some unknown period before that, despite every walkthrough in that window logging the standby as available.

What confirming actual readiness requires

The fix is a periodic exercise test, not a presence check, actually starting the standby unit and running it under a real or representative load for long enough to confirm it starts cleanly, reaches rated performance, and runs without abnormal symptoms, on a defined schedule rather than only when the duty unit happens to fail. For critical lift stations and process units, alternating duty and standby roles on a regular rotation accomplishes the same confirmation as a dedicated test, since the previously idle unit is now the one actually running and proving itself, while the previously duty unit becomes the newly idle standby due for its own eventual exercise. Either approach converts redundancy from an assumption resting on a presence check into a demonstrated, periodically re-confirmed fact.

Exercise testing also interacts directly with how a plant manages spare parts inventory: a standby unit discovered failed during a scheduled, low-stakes exercise test gives the maintenance team time to source a replacement part through normal channels, while the same failure discovered during an actual emergency failover means sourcing that part under maximum time pressure, often at a premium and sometimes against a lead time the plant cannot absorb without extended downtime. The exercise test does not just confirm readiness; it converts an unplanned, high-stakes failure into a planned, low-stakes one, which is a meaningfully different problem to be managing even when the underlying mechanical fault is identical, and a meaningfully smaller line item on whatever downtime actually costs a plant once a lift station is genuinely offline with no working backup.

If your plant's standby pumps, blowers, or other backup equipment are confirmed available by presence alone, with no periodic exercise under load, that gap is worth closing with Spans' operations and maintenance team before a primary failure reveals the backup was never actually going to work.