A useful APFC alarm investigation preserves the code and operating state, then traces measurement, settings and physical stage response instead of clearing evidence first.
- Capture the alarm context before reset or parameter changes destroy it.
- Troubleshooting should follow the chain from measurement to decision to output to physical response.
- Protective alarms should not be bypassed to make the display appear normal.
Preserve the event before changing the panel
The alarm code, timestamp and operating state are the initial evidence set, not administrative details.
Capture displayed voltage, current, kW, kVAr, power factor, active steps and recent alarm history. Record whether the supply came from utility, generator or another bus arrangement and whether the load was stable or changing.
Use the exact controller manual because identical words can represent different trigger logic across models. Avoid repeated resets until the evidence and any required safety response are complete.
Evidence basis[1] Schneider Electric[2] ABB
Preserve the event before changing the panel
The alarm code, timestamp and operating state are the initial evidence set, not administrative details.
Final ratings, protection, earthing, settings and interlocks follow approved project engineering.
Open full-size SVGTrace measurement, logic and output in order
A structured chain separates bad input data from a valid decision that failed to reach the electrical branch.
First test whether the CT and voltage reference produce credible direction and phase relationship. Then compare settings with the approved stage schedule. Finally, verify output indication, control wiring and the switching device response.
If an output operates but reactive power does not change as expected, investigate the physical branch, protection, discharge state and capacitor condition rather than increasing controller sensitivity.
Evidence basis[1] Schneider Electric[2] ABB[3] International Electrotechnical Commission
Trace measurement, logic and output in order
A structured chain separates bad input data from a valid decision that failed to reach the electrical branch.
Final ratings, protection, earthing, settings and interlocks follow approved project engineering.
Open full-size SVGReturn to service with a monitored hypothesis
A repair is not closed until the predicted system response has been observed without recreating the fault.
Document the confirmed cause or state the remaining uncertainty. Restore settings from a controlled baseline, observe the corrected stage under an authorised load, and retain post-action measurements.
Where the event depends on an operating state that cannot be recreated safely, label the closeout as conditional and define the monitoring evidence required at the next occurrence.
Evidence basis[1] Schneider Electric[2] ABB[3] International Electrotechnical Commission
Information that makes the next review more useful.
- 01
Photograph or export the exact alarm, time, measured values and active stage indications before reset.
- 02
Record the source, bus-coupler, generator and load state present when the event occurred.
- 03
Confirm controller model, firmware where available, wiring diagram and current settings revision.
- 04
Check CT signal, voltage reference, phase relationship and control-supply condition.
- 05
Verify output mapping and the electrical response of one approved stage.
- 06
Inspect switching, protection, discharge and capacitor branch evidence after safe isolation.
- 07
Close the event with cause confidence, corrective action and a monitored return-to-service plan.
Standards and technical references.
Links identify the source and scope; access to a complete standard may require purchase or organisational access.
Questions engineers and project teams ask.
Should an APFC controller alarm be reset before investigation?
Capture the code, time, values, active stages and system state first. Reset only through the approved operating procedure after any safety response is complete.
Why can an APFC panel show a poor power factor with healthy capacitors?
Possible causes include CT boundary or polarity, phase-reference mismatch, settings, output mapping, an unavailable switching branch or a load state outside the configured response.
