Anti-Islanding Protection in Solar Inverters

Your panels are in full sun. The grid goes out. Your rooftop system shuts down with it and the house goes dark. Nothing is broken. That is anti-islanding protection doing exactly what Indian regulation requires.

Anti-islanding protection forces a grid-connected solar inverter to stop energising the utility line once the grid disappears. Without it, your inverter would keep pushing power into a cable that a line worker has isolated and believes is dead. It is not a fault, not a warranty issue, and not something an installer can legitimately switch off.

India makes this mandatory. The Central Electricity Authority requires cessation within two seconds of an unintended island forming, and every grid-tied inverter sold here is type-tested against IS 16169:2019, the Indian adoption of IEC 62116:2014.

Below: how detection works, why the non-detection zone never closes, the verified CEA trip limits, why a hybrid can legally keep your lights on, and what to do about nuisance trips. Every standard number was checked against the issuing body.

Short version. Anti-islanding protection makes a grid-connected solar inverter stop feeding the utility network within seconds of losing the grid. Detection combines passive sensing of voltage and frequency with active perturbation of the inverter output. Indian rules require cessation within two seconds of an unintended island forming, under CEA Regulation 11(6)(e), with type testing to IS 16169:2019.

What an island is, and why it is dangerous

An island is a section of distribution network cut off from the utility supply but still live, energised by generation on that section. A breaker opened or a fuse blew, and a rooftop inverter keeps exporting into copper that should be dead.

The CEA regulations define an unintended island as a part of the electricity system that stays energised by distributed generation after isolation. Deliberate islanding exists in microgrid design as an approved arrangement; anti-islanding prevents the accidental version. Islands survive longer than intuition suggests: a 5 kW inverter feeding roughly 5 kW of local load has no reason to stop.

Two hazards follow. The first is shock risk. Crews work a dead line by isolating, testing, and earthing it. A rooftop inverter backfeeding through the distribution transformer puts 11 kV back on the high-voltage side, so a worker following correct procedure can still be exposed.

The second is out-of-phase reclosure. Indian feeders use automatic reclosers that restore supply after a transient fault, assuming the section is dead. A surviving island has drifted out of phase by then. Closing two unsynchronised sources produces a large transient torque and current, damaging motors, transformer windings, and inverters.

Passive detection methods and where they go blind

Passive methods watch the inverter’s own terminals for something abnormal. They add no distortion and no cost, so every inverter carries them. Their weakness is structural: if the grid’s departure does not move the measured quantity, the method sees nothing. After the grid opens, inverter real power must equal island real load, forcing voltage to a new value; reactive mismatch moves frequency. Small mismatches mean small movements.

ROCOF measures df/dt rather than f, so it reacts before frequency drifts far enough to hit a fixed threshold. That speed is also its problem: a large generator tripping elsewhere produces a genuine slope, and a sensitive ROCOF element then disconnects a healthy plant.

Vector shift watches the duration of each voltage cycle. At separation the load angle jumps, shortening or lengthening one cycle. Fast, no injection needed, and fooled by motor starts, capacitor switching, and nearby fault clearance.

Active detection methods and the power quality tradeoff

Active methods stop waiting and start pushing. The inverter perturbs its output in a way a stiff grid absorbs without reacting, then watches. An island cannot absorb it, so the disturbance grows until a conventional trip fires. Detection is bought with distortion, so vendors trade perturbation size against harmonics, flicker, and stability on weak networks.

Positive feedback is the key idea. Plain AFD pushes with fixed force and a stubborn load holds it in place. A positive-feedback scheme raises its push in proportion to how far frequency has moved, so the island runs away from stability instead of settling into it.

The non-detection zone, and why inverters mask each other

The non-detection zone, or NDZ, is the band of real and reactive power mismatch where a scheme fails to trip in time. It is a property of every method, not a product defect, and shrinking it for one load condition usually widens it for another. The arithmetic, using round numbers rather than field data:

Now put several inverters on one feeder. Each unit’s perturbation becomes a smaller share of total island power. Worse, positive-feedback schemes from different vendors use different gains and signs of drift, so one pushes frequency up while another pushes it down and the two partly cancel.

Certification does not cover this. Clause 6.1 of IEC 62116:2014 states the test uses an RLC load resonant at the nominal frequency and matched to a single unit under test. One inverter, one load. Your feeder is not that.

The standards that actually apply in India

Four documents matter and they do different jobs. People conflate them, then quote a trip time from a test standard that contains none.

IEC 62116:2014 scopes itself as “a test procedure to evaluate the performance of islanding prevention measures used with utility-interconnected PV systems”. That is the whole claim: a laboratory method, not a grid code.

Our guide to Indian inverter regulations and BIS compliance covers the registration chain; the BIS certification glossary entry explains what a CRS registration means.

IEEE 1547-2018, amended by IEEE 1547a-2020, says the same in Clause 8.1.1: detect, cease to energize, and trip within two seconds of formation. The NREL primer notes the operator may extend that to five seconds, catalogued on OSTI. India does not run on IEEE 1547, but the figure matches the CEA rule.

Trip thresholds and clearing times

These come from Regulation 11(6) of the CEA Connectivity Regulations, 2013 as amended in 2019. Treat them as the outer envelope, not your site settings.

Read the proviso on the voltage and frequency rows: the appropriate licensee “may prescribe a narrower range” for both. That is why two compliant installations in different states carry different settings.

In practice your DISCOM sets the grid profile, usually narrower, and some utilities require an external protection relay above a defined capacity. Confirm it before commissioning, not after. Shipped settings belong in product documentation; the Qbits datasheet library is where you identify which document covers the exact model.

Hybrid inverters, backup mode, and the transfer switch

A hybrid inverter can keep a house lit during an outage and still comply, because it is not energising the utility circuit. The requirement is precise: stop energising the electricity system, meaning the utility network. It says nothing about a circuit physically separated from that network first.

Inside a hybrid sits a transfer arrangement, usually relays or contactors, between grid input and backup output. On grid loss the sequence runs: detect, stop exporting, open the grid-side contacts, verify separation, then restart the backup output from the battery and available PV. It is break before make.

Two consequences. Backup loads see a real interruption, not a glitch-free handover. Qbits states UPS switching within 10 seconds for the relevant hybrid models; the figure is model-specific, so confirm it on that datasheet. Only loads wired to the backup output are covered, and a wall socket proves nothing about the changeover.

So both architectures stop energising the utility circuit inside the clearing time. Only the hybrid then opens its grid-side contacts and keeps designated backup loads running, within battery and output limits.

The on-grid versus hybrid comparison sets out the decision criteria, and the guide to inverter behaviour in a power cut covers what a homeowner experiences. Specifications sit on the hybrid inverter page.

How anti-islanding is verified at commissioning

Anti-islanding is not something you eyeball. Evidence lives in two places, a type-test report for the model and a site record for the installation, and neither substitutes for the other.

1. Identify the exact model and firmware on the nameplate and display, then match both to the test report. A family brochure may not cover every revision sold.
2. Collect the islanding test report against IS 16169:2019 or IEC 62116:2014, and confirm the edition and models covered.
3. Collect the separate safety report against IS 16221 (Part 2):2015, used with Part 1.
4. Confirm the loaded grid profile matches what the DISCOM approved, and photograph the settings screen.
5. Run the utility-witnessed disconnection test if required. Open the main incomer and record the time from loss of supply to cessation of output.
6. Confirm the reconnection delay, then record everything with date, meter readings, and names of those present.

IEC 62116:2014 calls the quantity in step 5 the run-on time: the interval between opening of the test switch and cessation of output current. On site you measure the same thing with cruder instruments. Our commissioning walkthrough covers the wider checklist.

Nuisance tripping on weak Indian feeders

Repeated trips on a site that otherwise generates fine are usually a network problem, not an inverter defect. Rural and semi-urban feeders run long, sit lightly loaded at midday, and hang off transformers with wide tap ranges.

Two mechanisms dominate. AC-side voltage rise comes first: the inverter exports through cable impedance, so terminal voltage exceeds the voltage at the pole transformer, and the gap grows with cable length and export current. The inverter sees an over-voltage a meter at the transformer would not show. Second, genuine frequency events, which ROCOF and vector shift react to whether or not an island exists.

The grid over-voltage guide gives the measurement sequence; the error code reference maps a displayed event to the right service route.

Five things anti-islanding does not do

The function does not exist to help the system owner. It protects people outside your property boundary, and does so by making your system less useful during an outage.

Myth 1: it protects your home during a power cut. It does the opposite. It is the reason a standard grid-tied system goes dark.

Myth 2: it can be switched off for backup. Disabling it violates the interconnection requirement and creates a live-line hazard for utility crews. The legitimate route is a hybrid with a transfer arrangement.

Myth 3: hybrid inverters do not have anti-islanding. They do, and they must. The transfer arrangement is what lets them supply a backup circuit safely.

Myth 4: it protects the inverter from surges and lightning. A different function entirely. Surge protection devices and earthing handle that.

Myth 5: a test certificate proves your feeder is safe. It proves one inverter passed a defined test against one resonant load. It says nothing about the seven other inverters on the same transformer. Nor does frequent tripping mean your unit is defective; on a weak feeder it is usually reporting a real condition.

The Bottom Line

Anti-islanding protection is a safety obligation the inverter carries for the utility. It senses the grid’s absence, probes actively when sensing is ambiguous, and shuts down inside a clearing time set by regulation. In India that is two seconds from formation of an unintended island, under Regulation 11(6)(e) of the CEA Connectivity Regulations, 2013. No method is perfect, which is why utilities layer their own screening on top of certification.

Three things to do next:

• Pull the islanding test report for the exact model and firmware on your site, confirm it cites IS 16169:2019 or IEC 62116:2014, and file it with the commissioning record.
• Confirm the grid profile loaded into the inverter matches what your DISCOM approved, and photograph the settings screen at handover.
• If you are specifying a system now, or need documented grid settings for a model, talk to the Qbits technical team before you finalise the interconnection application.