An inverter disconnects from the grid when one of its internal protections orders it or when the relay at the point of interconnection opens. The most common internal causes are grid voltage or frequency outside the range of the configured grid code, low insulation resistance in the PV array, high residual current, temperature, and active anti-islanding on weak grids. The inverter alarm list, cross-checked against the interconnection relay events, shows which one it was.
Symptom
One or several inverters in a solar plant stop injecting power and stay on standby, faulted or disconnected. It can happen on a single inverter, on all of them at once, at certain hours of the day or after rain. Sometimes the interconnection relay logs nothing; other times the disconnection came together with a trip of the plant breaker.
Two layers have to be separated, as explained in solar plant protection: the inverter, which has its own DC-side and AC-side protections and must stop energizing the grid when the grid goes out of range or disappears. The other layer is the point-of-interconnection relay, which protects the utility grid from the plant. If it opened the interconnection breaker, the inverters disconnect because they lost the grid: the cause is in the relay or in the grid. If the breaker is still closed and the inverters tripped, the cause is in the inverters or in what they see of the grid.
Possible causes, from most to least likely
| Cause | Why it happens | What gives it away |
|---|---|---|
| Voltage out of range | Voltage rise from export on a weak grid, sags from grid faults | Grid overvoltage or undervoltage alarm in the inverter |
| Wrong grid code | Profile from another country or standard, or ranges different from the interconnection relay | Disconnections on variations that should be ridden through |
| Low insulation resistance | Moisture, damaged connectors or cables in the PV array | Inverter does not start in the morning or after rain; insulation alarm |
| High residual current | Ground leakage on the DC side or array capacitance with moisture | Residual current alarm, sometimes intermittent |
| Temperature | Blocked ventilation, dirty filters, direct sun on the unit | Power derating and then disconnection in hot hours |
| Active anti-islanding on a weak grid | The active method reads the response of a weak grid as an island | Disconnections with no clear grid event, more frequent at high generation |
| Ride-through disabled or badly set | The inverter trips on sags it was supposed to ride through | Disconnection coinciding with a remote fault on the grid |
| Interconnection breaker opened | Interconnection relay trip, utility intertrip | Interconnection relay event before the inverters disconnect |
| Internal inverter fault | Power modules, boards, fans | Internal fault alarm, a single inverter affected |
What to check first
- Alarms of each inverter with exact time, and whether one or all are affected. A single inverter points to it or to its array; all at once points to the grid or the point of interconnection.
- Interconnection relay events at the same time. Also review the plant controller log and the requirements in protection for distributed generation that apply to that relay.
- Grid code profile loaded in the inverters and firmware version, compared with what was approved. In the Huawei SUN2000 inverter, for example, the profile is selected in the device configuration.
- Insulation resistance and residual current recorded by the inverter, and their relation to humidity and time of day.
- Internal temperature and ventilation condition.
- Did the point-of-interconnection breaker open?
- Yes: The cause is in the interconnection relay or the grid: review its events
- No: Did a single inverter trip or all of them?
- A single one: Which alarm did it log?
- Insulation or residual current: Insulation fault in its PV array
- Temperature or internal fault: Ventilation or the unit itself
- All of them: Did voltage or frequency leave the grid code range?
- Yes: Grid event or voltage rise: review the grid code and reactive power control
- No: Active anti-islanding on a weak grid or ride-through badly set
- A single one: Which alarm did it log?
How to confirm it
- Insulation measurement of the PV array string by string, to find the one that lowers the resistance. See DC ground fault detection.
- Voltage and frequency recording at the point of interconnection with an analyzer, compared with the ranges of the loaded profile.
- Review of the grid code profile against the requirements approved by the utility and against the interconnection relay settings. See voltage sags and grid support in renewables.
- Thermal inspection of the inverter and its connections.
How to keep it from happening again
- Grid profile verified at commissioning, with the certified model and firmware. Verify that they match those on the certificate. See commissioning tests of solar plants and, unit by unit, on-site inverter tests.
- Coordination between the inverter and the interconnection relay: the relay settings must stay outside the inverter ride-through curves.
- Tell the station trip from the inverter’s own: if the event appears first in the medium-voltage switchgear or the transformer, the guide on trips in a solar transformer station helps identify the origin.
- PV array maintenance: connectors, string boxes, cables and drainage.
- Cleaning of filters and ventilation as the manufacturer indicates.
- Voltage or reactive power control agreed with the utility if the grid is weak.
- Firmware under control: an update can change the inverter behavior or leave it with grid parameters different from the approved ones. See inverter firmware: safe updates and, for device access and configuration, the guidance in NIST IR 8498 and IEEE 1547.3.
What not to do
- Do not switch the grid code profile to a more permissive one to avoid disconnections: it must match what was approved.
- Do not disable anti-islanding in the inverter: it is a safety protection for people working on the grid.
- Do not ignore an insulation alarm that goes away when things dry out: the defect is still there and can end in a ground fault.
- Do not update the firmware without verifying that the new version is still covered by the certification.
What the standards require: when it must trip and when it must not
Before blaming the inverter it helps to know what is required of it. IEEE 1547-2018 does not require a distributed energy resource to disconnect on any disturbance: it defines abnormal-performance categories (I, II and III) that set the voltage and frequency ranges in which the plant must keep operating and those in which it may trip. Categories II and III demand more frequency response capability than category I, and the IEEE 1547a-2020 amendment widened the permitted clearing-time ranges for category III. In practice, a disconnection inside the “must remain connected” zone is a nonconformity; one outside it may be the expected behavior.
Two concepts from that standard help in reading the events:
- Trip is not the same as physical disconnection. The standard defines it as the cessation of energy injection without immediate return to service; the inverter contactor may still be closed. What follows is the return to service, which depends on the reconnection settings, and that is why a fault of a few seconds turns into minutes without generation.
- Unintentional islanding. This is the case where the resource energizes a part of the grid. IEEE 1547-2018 requires detecting it and ceasing to energize within a maximum of 2 s from its formation, according to utility requirement documents that cite it. The standard does not fix the method the manufacturer uses to achieve this. These measures are tested under IEC 62116, whose second edition dates from 2014. It describes a test method for single-phase and multi-phase grid-connected PV inverters.
For a utility or an auditor, the right question after a disconnection is which category and which profile the equipment is certified for, and whether the profile loaded in the field matches. That evidence, together with the model and firmware on the certificate, is what is reviewed in a commissioning test. If your plant is suspected of an interaction between anti-islanding and a weak grid, see also anti-islanding protection.
When to contact us
If the inverters in your plant disconnect with no clear cause, send us the inverter alarms and the interconnection relay events: we will quote the event analysis. We review the coordination between the point-of-interconnection relay and the inverters through remote relay configuration, and on site we do the testing and commissioning of solar plants, in Colombia or elsewhere on request. Write to us.
Related guides
- Solar plant protection
- Huawei SUN2000 inverter
- DC ground fault detection
- Commissioning tests of solar plants
- Trips in a solar station: switchgear, transformer and inverter
- On-site inverter tests: startup and grid functions
- Inverter firmware: safe updates
References
- IEEE 1547-2018, Standard for Interconnection and Interoperability of Distributed Energy Resources with Associated Electric Power Systems Interfaces.
- IEC 62116, test procedure of islanding prevention measures for utility-interconnected photovoltaic inverters.
- IEC 62109, safety of power converters for use in photovoltaic power systems.
Frequently asked questions
Why does the inverter not connect in the morning?
A frequent cause is the insulation resistance of the PV array. Before connecting, many inverters measure it; with moisture or dew on modules, or with damaged connectors or cables, the value drops below the threshold and the inverter waits. If it improves as things dry out, the problem is insulation on the DC side, not the inverter.
What is the inverter grid code?
It is the set of voltage, frequency, timing, ride-through and reconnection settings the inverter applies. Manufacturers offer profiles by country or by standard. If the loaded profile does not match what the site requires, or does not match the interconnection relay, the plant trips when it should not or stays on when it should trip.
How long does the inverter take to reconnect after a disconnection?
Whatever its reconnection setting says, which must match what was agreed with the utility. Many grid codes set an automatic reconnection delay of a few minutes (for example 3 to 5 minutes) after the grid returns to normal conditions. That is why a brief grid disturbance turns into several minutes without generation.