The word microgrid gets applied to any site with solar and a battery. The distinguishing feature is not the equipment. It is whether the site can deliberately disconnect from the grid, run independently, and reconnect without disruption.
Three capabilities:
Intentional islanding. The site can separate from the grid on command or automatically on detecting a problem, and continue supplying its loads.
Grid-forming generation. Something has to establish voltage and frequency once the grid reference is gone. Standard grid-following inverters cannot – they need an existing reference. This requires grid-forming inverters, a synchronous generator, or both.
Coordinated control. Something must decide, in real time, what runs, what sheds, how sources share load, and when to reconnect.
Solar and a battery without those is a solar-plus-storage installation. Useful, but it will go dark with the grid.
This trips up more projects than any other aspect. A conventional solar inverter, on losing the grid, disconnects – correctly, per anti-islanding requirements.
In an island you need a source establishing the reference. Usually the battery inverter operates in grid-forming mode, creating a stable voltage and frequency that the solar inverters then follow.
Which means the battery inverter must be specified as grid-forming capable. Not all are, and the difference is not always obvious from a datasheet headline. Ask explicitly whether the unit supports grid-forming operation and seamless transfer, and what the transfer time is.
For many critical facilities the concern is not a long outage but the transition itself.
Open transition disconnects, then starts island operation. There is a break – typically tens to hundreds of milliseconds. Fine for lighting and most motors; not fine for some process control and IT equipment.
Closed transition or seamless transfer maintains supply through the transition, with the battery inverter taking over without a measurable break.
The distinction determines whether you still need a UPS for sensitive loads. Many sites end up with a layered arrangement: UPS for the most critical equipment, microgrid for the facility, generator for extended outages. That is not redundant – each layer addresses a different duration and criticality.
In island mode you almost certainly cannot supply everything. Available generation is limited, and if load exceeds it the island collapses.
So the control system needs a defined priority hierarchy: which circuits are essential, which are deferrable, which shed immediately. This requires the electrical distribution to be arranged so those groups can be separated – which frequently means switchboard modification, and is the part that turns a controls project into a construction project.
Doing this properly requires the facility to make genuine decisions about priority. In my experience that conversation is harder than any technical aspect, because everyone believes their load is critical.
Returning to grid operation requires synchronisation – matching voltage, frequency and phase angle before closing. Closing out of synchronism can damage equipment severely.
Automatic synchronisation is standard in proper microgrid controllers. Verify it is included and tested, and that the system waits for a stable grid rather than reconnecting to a supply that is still fluctuating.
A microgrid that has never been tested under realistic conditions is an assumption.
Commissioning should include deliberate islanding under load, operation in island mode for a meaningful period, load shedding verification, and synchronised reconnection. Then repeat periodically – annually is reasonable – because configurations drift, loads change and firmware updates alter behaviour.
The facilities that get value from microgrids are the ones that test them. The ones that installed and never exercised the capability generally discover during a real outage that something changed three years ago.