Table of Contents

Distributed solar power generation is solar electricity made at or near the load it serves, or fed into the distribution network, rather than a central station built to feed the high-voltage transmission system. The useful test is the meter and the wires, not the kilowatt rating. Pacific Northwest National Laboratory’s hardware description already shows both ends: a small array, under about 50 kilowatts in that account, can land on a building service panel, and a larger array may need a transformer to step up toward grid voltage. Their example, 480 volts up to 15 kilovolts, illustrates “larger.” It is not a cutoff that every country uses, and it is not a rule that a 60-kilowatt roof has stopped being distributed. A warehouse roof of several hundred kilowatts is still distributed if it serves that site.

A statistical definition used in international reports, and the Chinese policy categories for plants on roofs, greenhouses, or open ground, are handled in distributed photovoltaic scenarios. This article stays with the electrical choice a buyer is actually making: who consumes the energy, what happens in an outage, and which equipment that implies.

On the customer side of the meter

Most of what people mean by a distributed system is behind the meter. The modules, the inverter, and the building share a service. Midday production reduces what the building buys. Surplus may export if the utility allows it, at a price that may not match the retail rate. The building’s evening load is not automatically served. Storage does that only when storage was added for that hour, which is a separate decision in what storage is for. Calling the roof “the power plant” skips the hour.

The inverter in this arrangement is still a grid device. IEEE Std 1547-2018 requires it to cease energizing an unintentional island within 2 seconds. The National Renewable Energy Laboratory primer describes that as the normal US listing path, typically through UL 1741, with ride-through and voltage support tested rather than treated as optional extras. Distributed, in other words, does not mean electrically casual. A larger share of distribution-connected inverters is why the standard added ride-through: the primer’s concern is a fleet that all trips at once. The development of that requirement is the subject of how photovoltaic inverters are developing. It applies to a house and to a factory roof.

Not every distributed system is behind a customer meter

A system can connect on the utility side of a customer’s meter and still be distribution-connected rather than a transmission plant. The buyer’s questions change. There may be no on-site load to absorb the noon peak. The value is the power delivered to the feeder, under a contract, not a lower retail bill. Curtailment, reactive power, and the ride-through settings become the product. A sales comparison that uses a home self-consumption spreadsheet for that plant is the wrong model.

Off-grid solar is easy to misfile as distributed because it is small and local. It is a different electrical object. There is no utility to cease energizing. The inverter has to form the AC system, and the battery has to cover the night. Classifying those systems by the load they serve is in independent photovoltaic systems. Putting them in a distributed-generation total because a spreadsheet has nowhere else to put them does not change the equipment.

What distributed generation is allowed to claim

It can claim production at the site, if the meter or the inverter log is real. Pacific Northwest National Laboratory wants that production compared with an expected month, and it treats about 10 percent variation in sunlight as ordinary. It can claim a lower purchase from the utility to the extent the load overlaps the solar day and the tariff says so. It cannot claim backup during an outage unless the islanding equipment exists. It cannot claim a safer roof merely because the inverter is listed. Modules remain live in daylight, and shutdown is still AC disconnect then DC disconnect, by trained people.

It also cannot claim that any sunny surface is an appropriate site. Ground, roof, and carport are PNNL’s three ordinary mounts, with different access and structure. A feeder that cannot accept the export is not a site, however good the roof is. Siting checks are in where a distributed system can be installed. Safety limits that survive a brochure are in whether distributed solar is safe.

A definition you can use in a specification

Write where the system connects, at what voltage, and whether it is allowed to export. Write whether it must ride through, and on whose settings file. Write whether any circuits must remain powered in a utility outage. Write the mount and who can reach it. If those sentences are missing, the phrase “distributed solar” has not specified a plant. It has specified a mood. The kilowatts come after the connection, not before.

Two sites with the same module count can still be different plants under that test. A 200-kilowatt roof that offsets a weekday factory load is customer-side distributed generation. A 200-kilowatt array that only exports onto a feeder, with no on-site load, is still distribution-connected, but the contract and the midday surplus are the product. A 20-kilowatt array with a battery and no utility is neither of those. It is off-grid, even if a brochure files it under the same heading because the panels look the same. The specification should say which of the three was bought. The scenarios article is where site types and policy labels are sorted. This test is only the electrical boundary.

Frequently asked questions

Is there a maximum size for distributed solar?

Not in the sense that would turn a large roof into a central station. PNNL distinguishes how a small array, under about 50 kilowatts in that description, may use a service panel, while a larger one may need a transformer. The example step-up voltage is illustrative. The connection point is the definition.

Does distributed solar work during a blackout?

Ordinary grid-tied distributed solar stops. IEEE 1547-2018 sets the cease-to-energize time at 2 seconds. Outage power is a separate, intentional-island design. Off-grid systems are a third case and are not grid-tied distributed generation.

Is distributed generation the same as rooftop solar?

Rooftop solar is one mount. PNNL also lists ground arrays and carports. A ground system serving a building or a feeder can be distributed. A roof system aimed only at transmission delivery would not fit the local-connection meaning used here.

Sources and references

How small and larger arrays connect, mounts, production checks, and live modules: Pacific Northwest National Laboratory. Cease-to-energize, ride-through, and why a fleet of distributed inverters is treated as grid equipment: NREL/Sandia IEEE 1547-2018 primer, 2022.