Microgrids for Large Sites: When They Make Sense and What to Connect First

What counts as a microgrid on a large site

The U.S. Department of Energy describes a microgrid as a group of interconnected loads and local energy resources within clearly defined electrical boundaries that acts as a single controllable entity with respect to the grid. Three attributes follow from that definition: the system can transition between grid-connected and islanded operation, it can draw on whatever localized resources are available, and an advanced controller orchestrates the balance between generation and demand. A large campus, industrial park, port, airport, hospital district, or military installation can all qualify if those boundaries and controls exist.

That definition quietly excludes a lot of equipment people call a microgrid. A standby diesel generator with a manual transfer switch is backup power, not a microgrid, because nothing coordinates multiple resources. A grid-tied solar array that shuts down when the utility drops is distributed generation, not a microgrid, because it cannot form an island. The distinction matters for planning: if your goal is continuity through an outage, the system must be able to separate from the grid, sustain its own voltage and frequency, and serve a defined set of loads on its own.

When a large-site microgrid makes sense

The clearest signal is expensive downtime. If an hour without power costs a facility in spoiled product, lost production, interrupted care, security gaps, or regulatory penalties, a microgrid becomes an economic decision rather than an environmental gesture. U.S. defense installations have pursued microgrids in part because DoD policy requires improved energy resilience across installations, and projects such as the microgrid at Marine Corps Air Station Miramar emerged from that requirement.

Other signals point the same way: high or volatile energy prices that make self-generation and storage attractive, a constrained or expensive grid connection that limits expansion, or a remote location where extending the utility network is impractical. A microgrid can also defer costly infrastructure upgrades by reducing peak demand on the wider grid.

• The site has a defined set of critical loads that must stay energized during an outage.
• Outage history or extreme-weather risk makes an extended interruption plausible.
• Fuel delivery for backup generators is uncertain or expensive.
• There is space and solar or wind resource for on-site renewable generation plus storage.

When it probably does not make sense

A microgrid is not automatically better because the territory is large. If an outage is merely inconvenient, if the site can shut down safely and restart cheaply, or if the local grid is already reliable and inexpensive, the upfront cost and engineering effort may never be recovered. The U.S. Department of Energy's own program notes that adoption is limited by high upfront costs, complex design, and regulatory barriers, which is a useful reality check for any business case.

Size can also work against you. The larger and more varied the territory, the harder it is to define clean electrical boundaries, coordinate protection settings, and maintain selectivity when fault currents change between grid-connected and islanded modes. Starting smaller, or staging a microgrid across one building or one process first, often reveals integration problems at low cost before the whole site is committed.

What to connect first: load triage before equipment

The single most valuable early task is to separate every load on the site into tiers, because the tier you protect determines almost everything downstream. A common three-tier scheme is critical, priority, and deferrable. Critical loads must run continuously through an outage; priority loads run when there is enough capacity; deferrable loads are shed first when generation is short.

Critical typically includes life-safety and process-safety systems, controls and monitoring, emergency lighting, communications, security, refrigeration for perishable stock, and any process that cannot be interrupted without damage. Priority often includes ordinary lighting, ventilation, office equipment, and pumps that can tolerate brief interruptions. Deferrable covers things like electric water heating, comfort cooling, EV charging, and decorative or outdoor lighting.

The triage has to be physical, not just a spreadsheet. If the boiler controls share a circuit with the breakroom kettle, the system cannot save one and drop the other. Priorities are embedded in the switchboard and cable layout, which is why load classification should be settled before equipment is specified.

• List every load, its normal running power, and its startup or inrush requirement.
• Mark which loads tolerate interruption, and for how long.
• Group loads onto separate feeders so critical circuits can be islanded cleanly.
• Record which loads can be throttled rather than fully disconnected.

Power versus energy: two separate sizing questions

After triage, sizing splits into two distinct numbers. Power, in kilowatts, is how much the critical tier draws at once, including motor starting surges. Energy, in kilowatt-hours, is how much the tier consumes over the outage duration you want to survive. Confusing the two is the classic planning error: a battery rated for the right power can still run out of energy hours too early, and a generator sized for average load can stall on a motor start.

There is also a control-layer question. In islanded mode, something must establish voltage and frequency for the island. A synchronous generator does this naturally; a grid-following solar inverter does not, because it needs an existing grid to synchronize to. A system that will island on renewables and batteries needs at least one grid-forming source, and multiple forming sources must be explicitly coordinated to avoid circulating currents and instability.

Finally, plan the transition. Seamless transfer is not guaranteed. Some hybrid inverters can switch a dedicated backup output in milliseconds, while a generator needs time to detect the outage, start, and take load. Loads that cannot tolerate even a brief dip should sit behind their own UPS, with the microgrid providing the longer-duration layer behind it.

From concept to build

Structured guidance for microgrid projects is generally phased: scoping and planning first, then data collection on loads, resources, and infrastructure, then conceptual design, then project development. The U.S. Department of Energy's free Microgrid Design Toolkit supports this by letting designers explore a large trade space of candidate designs and compare them on performance, reliability, and cost, surfacing the trade-offs of each choice.

Formal studies usually follow: protection coordination, power flow, short-circuit and arc-flash, and grid interconnection requirements with the utility. The goal of all of it is to prove, before procurement, that the island can separate, sustain itself, and reconnect safely. That proof is what separates a designed microgrid from an expensive collection of generators, panels, and batteries that happens to sit on the same site.