The Fault You Can't See Because You Can't Run It Long Enough

Why this matters

Some faults need time to appear, more time than a service call usually gives them. The component has to heat-soak, the tank has to fill, the drift has to accumulate, the system has to complete enough cycles. The customer says it acts up after an hour, and you have twenty minutes on site, so it runs perfectly the whole time you watch. Write that up as no fault found and you have failed a real fault and set up an angry callback. Recognizing a time-to-manifest fault, and knowing how to catch one you cannot sit and wait for, is the difference between an honest diagnosis and a shrug.

What a soak fault is

Time-to-manifest is the runtime a fault needs before it shows. These faults are gated by duration, not by load state and not by a single condition you can dial in on the spot:

• Thermal soak. A part that is fine cold but drifts, expands, or breaks down only after it has been hot for a while. Short runs never get it there.
• Fill, drain, or accumulation time. A level, a pressure, or a buildup that only crosses its trip point after the system has run long enough to get there.
• Duty-cycle wear. A fault that shows only after many starts and stops, not on the first clean cycle you watch.
• Slow drift. A setting, a reading, or an output that creeps out of range over hours, imperceptible in a snapshot.

Why a short visit is built to miss it

A brief on-site test samples the system at its best: freshly started, still cold, still in range. That is exactly the window before a soak fault wakes up. The clean twenty-minute run is not evidence of a healthy system; on a time-to-manifest fault it is the expected result and proves nothing. Treat a short passing run as inconclusive, not as a clearance, whenever the complaint is time-linked.

Estimate the time it needs, then match it

Ask the customer the most valuable question directly: how long does it run before it acts up. That number is your target. If it takes an hour, a fifteen-minute test cannot clear it, and you either commit to the full run or you use a workaround. Pin the time-to-manifest before you decide the visit found nothing.

When you cannot sit and wait

You rarely get to idle on site for the full soak, so get the evidence another way:

• Accelerate the condition, safely. Push the system toward the trigger faster where it is safe, more runtime, higher demand, a warmer confined space. Safety branch first: never accelerate heat, pressure, or electrical load past a rating or defeat a limit to force a fault faster, verify guards and cutoffs first, and stop if you are pushing the system outside its safe envelope.
• Trend a proxy reading. Watch a value that moves toward the fault, a temperature climbing, a pressure creeping, a current rising, and read the trajectory. You can often see it heading for the trip point well before it gets there.
• Leave a data logger. A recording instrument left in place captures the fault at the hour it actually happens, on the customer's real duty, without you present.
• Enlist the customer. Ask them to note the time, the runtime, and the conditions when it shows, or to capture it. Their record of the real event beats your snapshot of a healthy moment.
• Schedule a long-run test. Book time specifically to run the system to soak, or return at the hour the customer says it fails.

Do not launder a short run into a clearance

The failure mode is writing no fault found because the fault did not show in the time you had. That is not the same statement, and the customer knows it when the problem returns that afternoon. Say what you actually observed: the system ran clean for the time tested, the complaint is time-linked, and confirming it needs a full-duration run or a logged capture. Honest and inconclusive protects you; a false clearance does not.

References

• Trade-standard practice for extended-run and soak testing
• Manufacturer specified run-time and duty-cycle test procedures
• See related: The Tells That a Fault Is Condition-Dependent; It Worked When You Tested It: Why It Failed After You Left