An SF6 circuit breaker failure rarely announces itself. Most of the time the first indication is a breaker that will not close when the operator asks, or a low-gas alarm that appeared at 2 a.m., or a trip that never happened and a feeder that stayed live into a fault. The troubleshooting skill that matters is not knowing every possible fault — it is knowing which family of fault you are in within the first ten minutes.
This guide works through the failure modes of outdoor high-voltage SF6 breakers from 40.5 kV to 252 kV, the class represented by the high voltage SF6 circuit breakers range. It is written for the engineer standing at the control cabinet with a multimeter, not for the design office. Safety comes first throughout: nothing in this article overrides your switching authority’s procedures.
Almost every SF6 breaker fault shows up as one of five observable symptoms. Identifying the symptom correctly is half the diagnosis, because each symptom narrows the fault family immediately.
| Symptom | What the operator sees | Most likely families |
|---|---|---|
| Will not trip | Close works, open command produces nothing or a partial operation | Control circuit, trip coil, latch, density lockout |
| Will not close | Open works, close command rejected or spring does not recharge | Spring/motor, close coil, interlock, synchrocheck |
| Trips without command | Unexplained trip, often with no protection flag | DC earth fault, vibration, auxiliary contact, EMI |
| Gas alarm or lockout | Low density alarm, then lockout, breaker blocked | Leak, seal, density relay, gauge |
| Overheating or flashover | Hot terminal, discolouration, tracking, audible discharge | Contact resistance, loose joint, polluted insulator, moisture |
Two of these five — gas alarms and overheating — are progressive and give warning. The other three are binary and appear without notice, which is why they dominate unplanned outages.
Every fault belongs to one of four subsystems. Sorting the symptom into a family is the fastest way to stop guessing.
| Family | Typical causes | Share of failures |
|---|---|---|
| Gas system | Seal leaks, gauge root leaks, burst disc seepage, moisture ingress, density relay drift | Common, usually slow |
| Operating mechanism | Spring fatigue, damper wear, latch wear, motor failure, lubrication loss, linkage wear | Most frequent cause of refusal to operate |
| Contact and insulation | Contact erosion, loose terminal, polluted porcelain, moisture, partial discharge | Less frequent, highest consequence |
| Control and secondary | DC supply loss, coil open circuit, wiring termination, auxiliary contact, interlock, EMI | Most frequent cause of “breaker won’t move” |
The distribution matters. In practice, a breaker that refuses to operate is far more likely to have a control or mechanism problem than an interrupter problem — which is why the first test is always electrical, not mechanical.
A breaker that will not trip is the most dangerous failure in the list, because the backup is breaker-failure protection tripping the surrounding bays. Work through it in this order.
One recurring case deserves its own line: the breaker that trips correctly on a local test but not from the relay. That is almost always an interposing relay, a wiring termination, a fuse in the trip circuit, or a protection output card — never the breaker. Test the whole trip path, from relay output to coil, and record the result.
Closing failures are usually more benign but equally disruptive, because a bay that cannot be restored delays everything downstream.
The distinction to keep in mind: a close failure with a healthy motor and a charged spring is almost always electrical or interlock. A close failure with an uncharged spring is almost always mechanical.
The density relay is the breaker’s own protection against itself, and it produces more nuisance investigations than any other device on the unit.
First, verify the reading. Density relays drift and gauges fail. Cross-check with a calibrated pressure gauge and apply the temperature correction from the manual. A reading taken at 5 °C in the morning and one taken at 35 °C in the afternoon are not comparable unless corrected. A genuine low-density condition will be confirmed by both the relay and the corrected gauge reading.
Second, distinguish alarm from lockout. The alarm stage typically sits a few percent below rated density; the lockout stage sits below that. If the breaker is in lockout, it will refuse to operate — and that refusal is correct behaviour, not a fault to be bypassed. Never defeat a density lockout to restore service.
Third, localise the leak. The common leak points on an outdoor SF6 breaker are predictable: the density relay and gauge roots, flange gaskets between interrupter and support column, the burst disc, the operating rod seal at the mechanism entry, welded seams on the tank, and the sealing washers on any test or filling valve. A leak survey with a portable detector around these points will find the majority within an hour.
Fourth, quantify it. If the detector is inconclusive, bag the suspect joint and measure the concentration build-up over a known interval. A quantitative leak rate tells you whether you can defer the repair to the next planned outage or must act now.
One case looks like a leak but is not: a breaker that alarms in cold weather and recovers by afternoon. That is a density relay set too close to the operating point, or a gauge that is not temperature compensated. It still needs correction, but it is a relay problem rather than a gas problem.
Three methods cover nearly every situation, and they are complementary rather than alternatives.
| Method | Best for | Limitation |
|---|---|---|
| Portable sniffer detector | Localising an active leak on accessible joints | Wind disperses gas; needs slow, close passes |
| Bagging and concentration build-up | Quantifying a suspected joint leak | Requires the joint to be baggable and time on site |
| Ultrasonic detection | Very small leaks and pressurised compartments | Background noise in live substations |
Whatever method you use, record the date, the compartment, the density before and after any top-up, and the quantity added. Without those records, the next engineer inherits no history and starts from zero.
These are the failures with the highest consequence, because they end in flashover or fire rather than a refusal to operate.
Overheating at terminals. A loose connection raises resistance, which raises temperature, which further degrades the connection. Infrared survey under load finds it early. Left alone it progresses to a melted terminal pad or a burnt bushing flange.
Contact erosion inside the interrupter. Shows up as rising static contact resistance and as an accumulated breaking duty beyond the interrupter’s rating. It does not show up visually on a sealed unit, which is why the resistance trend matters.
Polluted or damaged insulators. Coastal salt, industrial dust and cement plant emissions all deposit conductive layers on porcelain. In wet conditions the surface tracks and eventually flashes over. Creepage distance specified for the wrong pollution level is the root cause, and no amount of cleaning fixes a fundamentally undersized insulator.
Moisture inside the gas compartment. Excess moisture lowers the dielectric strength of SF6 and, combined with arc decomposition products, forms corrosive compounds that attack internal surfaces. Rising dew point between visits is the warning; a gas dry-out or replacement is the remedy.
Partial discharge. A sustained discharge inside a compartment indicates a metallic protrusion, a loose shield or a floating potential. It degrades progressively and should be treated as a scheduled intervention, not a monitoring exercise.
Mechanism faults produce the classic signature of a breaker that operates, but not properly. Slow opening, incomplete travel, a loud closing bang, or a breaker that operates once and then will not recharge.
Comparing operating times against the commissioning baseline is the single most effective mechanism diagnostic. A change of a few milliseconds is noise; a 20% change in opening time is a mechanism that needs attention now.
A trip with no protection flag is one of the harder faults to pin down, and it is almost always in the secondary system.
A consistent sequence prevents the two most common field errors: chasing the wrong subsystem, and disturbing a healthy circuit while looking for a fault.
Record every measurement with its date and ambient conditions. Half of all repeat failures are the same failure returning because the first diagnosis was never written down.
| Situation | Usual decision | Reason |
|---|---|---|
| Single seal leak, compartment otherwise sound | Repair the seal | Cheapest correct fix; restore density and re-test |
| Repeated leaks on the same compartment | Replace the compartment or pole unit | The sealing system has failed structurally, not locally |
| Density relay or gauge fault | Replace the instrument | Gas system is intact; do not open the compartment |
| Trip or close coil failure | Replace the coil | Component-level repair, no gas work required |
| Mechanism wear after many operations | Mechanism overhaul | Restores timing and reliability at a fraction of replacement cost |
| Contact resistance trending up on all poles | Assess interrupter duty, plan pole replacement | Wear is internal and cumulative |
| Flashover damage to interrupter or bushing | Replace the pole unit | Insulation damage is not repairable in the field |
| Breaker beyond rated operations and past overhaul life | Replace | Every subsystem is at end of life simultaneously |
For 252 kV installations where the breaker is a major capital item, the decision is worth modelling properly. Units built for 50 kA breaking at 252 kV with 4000 A continuous rating are usually worth refurbishing rather than replacing, provided the interrupter itself is sound. The insulation margin that makes that judgement possible is visible in the type-test figures — 460 kV power-frequency withstand and 1050 kV lightning impulse withstand to earth at this voltage class.
SF6 itself is not toxic, but the gases formed when an arc decomposes it are. Sulphur dioxide, sulphuryl fluoride, thionyl fluoride and hydrogen fluoride are all present in arced gas, and they attack lungs, eyes and skin.
The pattern across all of these failure modes is consistent. Most breaker failures are electrical or mechanical, not gas-related, and most of them can be identified with a multimeter, an event log and a comparison against commissioning values. The interrupter, the most expensive component, is also the one that fails least often.
That is why the sequence matters more than the checklist: read the log, verify the DC, measure the coils, confirm the interlocks, then move to gas and mechanism. Following that order finds the fault in minutes instead of hours, and it avoids disturbing a healthy gas compartment in search of an electrical problem.
Where the diagnosis points to wear rather than fault, the maintenance intervals and checklists are the right next read. Where it points to a component that has failed its acceptance criteria, the testing regime defines what “failed” means, and the underlying limits come from the IEC 62271-100 and IEEE C37 series.
Secondary and control circuit problems — DC supply loss, open trip coils, wiring terminations and interlocks — followed by operating mechanism wear. Interrupter failures are comparatively rare but have the highest consequence.
Because pressure falls with temperature while density does not. If the alarm clears as the day warms, the density relay is probably set too close to the operating point or the gauge is not temperature compensated. Verify with a corrected density reading before adding gas.
It should not be. Below the lockout level the interrupter may not be able to extinguish an arc, and attempting an operation can destroy the breaker. The lockout exists to prevent that attempt and must never be bypassed.
A DC earth fault, a worn latch combined with vibration, auxiliary contact bounce, EMI on unshielded control wiring, or a genuine protection operation with a misconfigured or masked flag. Read the disturbance record before suspecting the breaker.
Yes. Arced SF6 contains decomposition products including sulphur dioxide and hydrogen fluoride that are toxic and corrosive. Test the gas, wear the specified PPE, recover rather than vent, and treat the recovered gas as hazardous waste.
A control or interlock fault can be identified in under an hour with a multimeter and the event log. A gas leak survey typically takes one to two hours on site. Mechanism faults that require timing tests and a partial strip-down are usually a planned outage activity.
