The Normal Behaviors Customers Report as Faults

Why this matters

A large share of diagnostic calls end with nothing wrong. The equipment made a noise, dripped water, cycled oddly, or smelled like something, and the owner reported a fault that was the machine working exactly as designed. Those calls are not free: they consume a diagnostic slot, they put a truck on the road, and they end with a conversation that is awkward whichever way you handle it.

The trap is that every normal behavior here has a near-twin that is a real fault, often a serious one, and a tech who learns to wave these off will eventually wave off the twin. The skill is not recognizing normal, it is knowing the discriminator that separates it from its twin and never declaring one without the other.

The three you never call normal remotely

Before any of this: there are reports where "that is probably normal" is never an acceptable phone answer, no matter how confident you are or how many times that customer has called about nothing.

• Gas or fuel odor. Leave, do not operate switches, call the gas supplier or emergency service from outside. There is no benign explanation worth the risk of being wrong once.
• Burning, hot-plastic, or electrical odor that is not a first-run burn-off you can positively confirm. Shut off at the breaker if it can be reached safely, and get eyes on it.
• Water at or near an electrical enclosure. De-energize from a safe point and keep people away. Normal condensation and a live enclosure are not a combination anyone assesses over the phone.

Everything below assumes you have already cleared these three.

The rule worth keeping

Never confirm normal without naming the discriminator that would have made it abnormal. Say both halves to the customer, every time: "That is the expansion noise, it is normal. What would not be normal is the same sound repeating every time the valve operates. If you hear that, call me."

That closes the call honestly, teaches the customer a test they can run themselves, and protects you, because you have documented what you ruled out rather than just said it was fine.

Thermal expansion and contraction

The mechanism. Metal grows and shrinks with temperature. A run of pipe, duct, or panel restrained at a strap, hanger, or penetration builds stress until it releases, and that release is a tick, a crack, or a bang. It correlates with temperature change, not with pressure or flow.

Its twin. Hydraulic shock, a loose mount, or a component striking something, all of which produce a similar bang.

The discriminator. Timing and repeatability. Expansion noise happens during the temperature ramp, appears once or a few times as the run heats or cools, and cannot be reproduced on demand. Shock and mechanical striking are triggered by an event, usually a valve or a start, and reproduce nearly every time you trigger it.

Condensation and water that belongs

The mechanism. Any surface below the dew point of the surrounding air collects water. Cold lines sweat, cooling equipment produces condensate by design, and combustion in some appliance types produces water as a product.

Its twin. A leak, and this one is dangerous to dismiss because a slow leak and heavy condensation look identical at the puddle.

The discriminator. Trace upward to the highest point of wetting, as you would a corrosion stain. Condensation wets a surface evenly and beads; a leak has a point origin and runs from it. Then check whether the system loses charge, pressure, or level over time. Condensation costs the system nothing, a leak shows up as loss somewhere.

Cycling that looks wrong

The mechanism. Many systems are designed to cycle, and a correctly sized system on a light load cycles more often, not less. Some equipment also runs deliberate low-output or staged operation that a customer reads as "it never shuts off."

Its twin. Short cycling from a real fault: restricted flow, a control fighting itself, a failing protective device resetting, or an oversized system.

The discriminator. Look at cycle length against load rather than at cycle count. Normal light-load cycling completes the cycle with no lockout or reset behavior. Fault-driven short cycling terminates early, often on a protection input, and worsens as load increases. Ask the load question directly: is it doing this on the easy days or the hard days? Frequent cycling on easy days is usually normal, on hard days usually not.

Pressure relief, purge, and drain events

The mechanism. Equipment deliberately vents, purges, drains, or discharges as part of a cycle. Regeneration, defrost, condensate pump-outs, and scheduled blowdowns all produce sudden noise, water, or vapor when the owner is not expecting it.

Its twin. A genuine relief-device discharge caused by an over-pressure or over-temperature condition, which is a serious finding.

The discriminator. Scheduled versus provoked. A designed purge or regeneration runs on a timer or counter, so it repeats predictably and stops on its own. A safety relief is provoked by a condition, so it correlates with load and recurs under that load until the condition is fixed. A reported discharge from a relief device specifically is not a normal-behavior call, it is a shutdown-and-inspect call.

Startup surge, inrush, and momentary effects

The mechanism. Motors and compressors draw far more current at start than while running, for a fraction of a second. On a soft supply that dims a light or clicks a relay elsewhere in the building.

Its twin. A failing start component, a degraded supply connection, or an under-sized or shared circuit under real stress.

The discriminator. Duration and trend. Normal inrush is momentary and its effect ends the instant the motor is up to speed. A fault-driven version lasts longer, worsens over weeks, or comes with the motor struggling, humming, or tripping. The customer-usable test: does the light dim for an instant, or stay dim for a second or more while the motor labors? That one distinction sorts most of these before anyone drives out.

Odors that belong

The mechanism. First-run burn-off of dust and manufacturing residue on hot surfaces produces a real burning smell that is genuinely normal and genuinely alarming. New materials off-gas. Seasonal first-run of equipment that has sat idle produces both.

Its twin. An overheating component, insulation breaking down, or an electrical fault. These smell similar enough that neither the customer nor anyone on the phone can tell them apart.

The discriminator. Duration plus context. Burn-off diminishes over the first runs and is gone within a few cycles, while a fault odor persists, strengthens with run time, or returns every run indefinitely. First run of the season on idle equipment is a plausible burn-off; the same odor mid-season with no idle period has no benign explanation and gets an inspection.

The field card

Worked example: the bang that was not shock

Complaint: a loud bang, once or twice, shortly after the system starts. The customer believes something is striking.

The call is worth taking seriously: hydraulic shock is real, damaging, and sounds identical over the phone.

On site, the test is built around the discriminator rather than around listening. Run the system through a full cycle from cold and note when the noise occurs relative to start. Then operate the valve or switching event repeatedly and see whether it can be provoked.

Results from a 20-minute observation run: the noise occurred once, roughly 2 to 4 minutes after start, during the fastest part of the temperature ramp. Three subsequent valve operations produced no noise. The event correlates with the temperature ramp and not with the valve, the exact inverse of the shock signature.

Physical confirmation: a run of metal restrained tight at a strap, with a bright rub mark showing movement. Loosening the restraint and adding an isolating pad eliminated the noise on the following cold-start run, verified before leaving.

The order matters. Provoking the valve is the cheaper test and it discriminates cleanly, so it runs before any teardown. A tech who started by inspecting the pipework would have found the rub mark and might have called it, but without the negative valve test the shock theory stays alive and the customer's real question, whether this is damaging the system, goes unanswered.

What would have changed the call: if the valve operations had reproduced the bang, this is shock, and the work moves to why - closure speed, an absent or waterlogged arrestor, or velocity in the run. If the noise had occurred at random with no correlation to either temperature ramp or valve, the next suspect is a mount or a component free to move, and the test becomes a run with the panel open and a listening tool at the mounts.

Verify you got this right

• Did you reproduce the customer's condition, not a convenient substitute? A cold-start behavior needs a cold start. Testing a warm machine and finding nothing is not evidence.
• Did you run the negative test as well as the positive one? Showing that the suspected trigger does not produce the noise is what closes a normal-behavior call properly.
• Did the customer leave with the discriminator? Not "it is normal," but "here is what would mean it is not."
• Is the ruled-out list in the record? A later failure of the twin is far easier to defend when the record shows you tested for it.

What changes the answer

• A behavior that changed. The strongest single signal here. A machine that has always ticked and still ticks is different from one that started last month, even if the sound is identical. Onset outranks character.
• A new installation. First-season noises, odors, and condensate are more likely to be genuinely normal, and more likely to be a commissioning issue than a wear fault.
• Occupied commercial or multi-tenant space. A normal behavior that disturbs occupants is still a problem to solve. Mitigation is legitimate work; do not close it as no-fault-found and leave.
• A customer who reports frequently. Resist the pattern-match. The discriminator still gets run every time, because the one call in the series that is real will arrive sounding exactly like the others.

The failure modes

Declaring normal from the phone. Every entry in the table above requires an observation the customer cannot reliably make unprompted.

Closing with "it is normal" and nothing else. The customer learned nothing, will call again about the same thing, and has no way to recognize the twin.

No record of what was ruled out. A no-fault-found visit with an empty record is indistinguishable from a visit where nothing was checked.

References

• Manufacturer documentation for designed cycle behavior, purge and defrost operation, and expected first-run characteristics
• OSHA and gas-supplier guidance on response to fuel gas odor and suspected electrical hazards
• Trade-standard practice for documenting ruled-out causes on a no-fault-found service call
• See related: Teaching the Difference Between Normal and a Problem; Teaching the Customer What a Normal Sound or Smell Is