AC-coupled / DC-coupled

What it is (precise)

The distinction is the electrical node where storage and generation meet, and it is the first thing a One-line diagram reveals. AC-coupled storage converts DC battery power to AC in its own PCS, steps it up through a dedicated MV transformer, and ties into the same AC collection system that the PV inverters feed and that the project's main power transformer carries onward to the grid.

Each subsystem stays electrically independent up to the AC bus and onward to the point of interconnection (POI), with its own protection, metering, and grid-code controls. Read the diagram: two inverter fleets on the MV bus means AC-coupled.

DC-coupled storage removes one conversion stage on the charge-from-PV path. The PV strings and the battery racks both land on a shared DC bus — commonly at or below 1500 VDC — behind one large inverter, with a DC-DC converter between the battery and the bus because the battery's VDC window rarely matches the PV array's operating voltage. Energy moving from panels into the battery never leaves DC, the architecture's defining efficiency advantage. The battery container/enclosure itself is largely identical in both cases; what changes is the conversion equipment wrapped around it.

The DC-DC converter is what makes the shared bus workable, and it is a four-quadrant step-up/step-down stage with two independent voltage windows, not a passive tap. One shipping converter accepts 550-1500 V on both its PV port and its battery port; another accepts 850-1500 V on both.

Because the windows are independent, the inverter can keep the array at its maximum-power point while the pack floats where its state of charge puts it — for a 1500 V LFP string that is roughly 1,150-1,330 V nominal, sagging toward a protected floor near 900-1,040 V, a cold and loaded figure rather than a sum of open-circuit voltages.

The converter also stands between the pack and any DC fault: one vendor describes the stage explicitly as limiting the battery's short-circuit current, which is one reason the DC-side protection study on a DC-coupled plant is not the same study as on an AC-coupled one.

Why it matters in a real grid-scale project

The choice drives round-trip efficiency (RTE) and how much energy you can actually deliver. DC-coupling avoids the DC-AC-DC double conversion when charging from PV — panel DC inverted to AC only to be rectified straight back to battery DC — and lets you recover otherwise-clipped energy — power the panels produce above the inverter's AC rating, which an AC-coupled plant simply spills.

On PV plants with DC-to-AC ratios of roughly 1.3 and above, clipping recovery can add several percent of annual energy, which is why DC-coupling is pitched hardest where the inverter rating and the interconnection limit are the binding constraints. Below about 1.2, the case thins out fast.

The trade is flexibility and operational independence. AC-coupling lets you size, site, dispatch, and even retrofit storage separately from the PV, route it through its own protection, and charge from or discharge to the grid when the solar is offline.

DC-coupling ties the assets together: a shared-inverter outage takes both down, the energy-management controls are more complex, and augmentation is harder because new racks must fit the existing DC bus. In the United States, investment-tax-credit rules on charging sources historically pushed developers toward DC-coupling; since standalone storage became ITC-eligible that pressure eased — check the rules current at your project date, they move.

Sharing one inverter is a rating constraint before it is a reliability one. Whatever the PV is producing at a given instant, the battery gets what is left of the inverter's AC rating and nothing more, so a DC-coupled block cannot discharge at full nameplate power through a sunny afternoon peak unless the inverter was sized for both at once — the opposite of the oversized-array logic that justifies DC-coupling in the first place.

Converter count is bounded too: one vendor caps parallel DC-DC units at six per inverter, so battery-side power on that block cannot exceed six converter ratings however many racks you buy. Protection is shared on the same terms. On that vendor's converter the PV-side fusing sits inside the inverter and ground-fault and insulation monitoring are the inverter's functions, while battery-side disconnection stays with the battery system's own breaker — one protection set covering two assets that an AC-coupled plant would protect separately.

Typical values and standards

Work the efficiency numbers on a consistent basis. A utility-scale PCS is typically 98-99% efficient one-way near rated load, so a full AC-DC-AC cycle costs roughly two conversion passes and whole-plant RTE lands well below the one-way figure.

DC-coupling's saving from skipping one conversion on the PV-charge path is on the order of one to two percentage points per avoided stage, before any clipping recovery — and the DC-DC converter you add back costs roughly a point itself, since shipping converters publish peak efficiencies around 99.0-99.2% but weighted averages nearer 98.2%, so budget the average rather than the headline.

That is why the net saving is about one stage, not two. Whole-plant AC-AC RTE at the POI typically sits around 85-92% for either architecture once transformer, cabling, and Auxiliary load — with HVAC the largest single consumer — are counted. At near-unity DC-to-AC ratios the DC-coupled edge shrinks toward the single-stage saving alone.

Hardware sizes cluster tightly. PCS blocks run roughly 1-5 MVA each — sized in apparent power, not MW — and DC buses sit at or below 1500 VDC, with 2000 V designs emerging. Safety and fire codes apply regardless of coupling, because the hazard lives in the lithium-ion cells, not the topology: fire propagation from thermal runaway is characterized by the UL 9540A test method, the assembled system is certified to UL 9540, and installation follows NFPA 855, which references NFPA 68 for deflagration venting and NFPA 69 for explosion prevention.

Inverters are typically listed to UL 1741; distribution-level interconnections reference IEEE 1547 and transmission-connected plants IEEE 2800; LFP dominates in both architectures.

A DC-DC converter's rating is a current limit wearing voltage clothing, exactly as a PCS rating is. One shipping 1500 V converter publishes 500 kW at 1000 VDC and 600 kW from 1200 to 1500 VDC at 30 °C — the same ±500 A continuous limit setting both rows, 500 A at 1000 V being 500 kW and 500 A at 1200 V being 600 kW, after which a power cap takes over and the rating stays flat to 1500 V while the current falls.

The same hardware derates on the axes a PCS does: another shipping converter is rated 1,200 kW at 30 °C, 1,120 kW at 40 °C and 1,040 kW at 50 °C, and derates above 2,000 m altitude at about 1.68% per 100 m, reaching roughly 66% of rating by 4,000 m. Listings follow the inverter's: both units above carry UL 1741 for North America and IEC 62109 internationally.

How it shows up in specs, studies and contracts

In an interconnection study the coupling choice sets what the POI actually sees. An AC-coupled hybrid presents two inverter fleets whose combined output must be modeled and often capped at the interconnection limit; a DC-coupled plant presents one fleet whose inverter rating is the hard ceiling for PV and battery combined.

Ask the grid operator whether the project registers as one resource or two — market rules for co-located versus hybrid configurations differ between ISOs such as ERCOT and CAISO, and that registration decides how you bid and get paid. Trace the single point of failure each one-line diagram implies before you sign anything.

On datasheets, check the DC-DC converter's rating and the battery VDC window against the shared-bus design in a DC-coupled plant: a converter whose minimum input voltage sits above the pack's protected floor strands usable energy at the bottom of the discharge, and does so without raising an alarm.

In warranties and capacity tests, pin the measurement basis: an RTE guaranteed DC-DC at the battery terminals runs several points above the same plant measured AC-AC at the POI, so a DC-coupled vendor quoting DC-side numbers is not comparable to an AC-coupled vendor quoting POI numbers. Force both onto the POI basis before comparing. Energy-yield models should state their clipping-recovery assumption explicitly — it carries most of the DC-coupled business case.

Metering is where the two architectures diverge most sharply on paper. An AC-coupled hybrid can carry a revenue meter on each subsystem's MV feeder, so PV output and battery throughput are separately measured quantities, each with its own accuracy class. A DC-coupled plant has no AC node between array and pack, so the only revenue-class measurement is at the POI and everything upstream of it is instrumentation.

What the converter reports is useful but coarser: one shipping DC-DC unit specifies power and energy measurement to better than 1.5%, improving to better than 0.5% with an optional high-accuracy DC metering kit whose battery-side measurement port is standard and whose inverter-side port is an option.

Those are instrument accuracies on a DC quantity, not meter classes — IEC 62053-22 classes 0.2S and 0.5S and ANSI C12.20 classes 0.2 and 0.5 are AC revenue-meter classifications and do not transfer to a DC bus. Settle the boundary in the contract: which meter the offtake settles on, which instrument any charging-source accounting relies on, and who owns the reconciliation when the two disagree.

Common pitfalls

The most common modeling error is comparing architectures on mismatched bases — DC ratings against AC ratings, or battery-terminal efficiency against POI efficiency. The number always looks better on the DC side, so insist every quote lands AC-AC at the POI.

The second error is assuming clipping recovery scales without limit: the battery can only absorb clipped energy when it has SOC headroom, so a system already full at midday recovers nothing, and the realized gain depends on dispatch strategy and state-of-charge management, not just the DC-to-AC ratio. Ask to see the hour-by-hour yield model, not a single annual percentage.

Retrofit and augmentation deserve early attention. Adding storage to an existing PV plant is almost always AC-coupled, because reworking the DC field is invasive; conversely, augmenting a DC-coupled plant later means sourcing racks compatible with the original bus voltage and converter design, sometimes years after the vendor has moved on. If long-term augmentation flexibility matters more than a few points of charge-path efficiency, that consideration alone can decide the architecture — and it is far cheaper to settle at the one-line stage than after the DC bus is built.