AC block
An AC block is the repeatable unit a grid-scale plant is bought and built in, and what defines it is where it stops: at an AC terminal. The common shape wraps some number of battery enclosures, the power conversion system serving them and the LV/MV step-up transformer into one engineered package carrying a single nameplate in apparent power — catalogue medium-voltage stations of that kind cluster between roughly 1.5 and 5.3 MVA per transformer.
The plant is then N copies of that block strung onto a medium-voltage collection run rather than a one-off design. Blocking buys one set of drawings, one factory test procedure, one spares list and one fault-isolation boundary; what it does not buy is a contract boundary, because the guarantee and the capacity test still land at the point of interconnection.
Reviewed August 2026 by Sergey Syrvachev
New to BESS? Start free with the 7-email fundamentals course — no cost, no account.
What it is (precise)
Two usages circulate and they are not the same width. Vendors selling integrated storage products use AC block for a battery-plus-inverter unit whose terminals are grid-ready AC, with the DC bus internal and outside the buyer's scope. Plant designers use it for the repeat unit of the layout: a group of enclosures, their conversion, the step-up transformer and the MV switching that ties the group onto a collection feeder.
The first is a product, the second is a plant module, and the block named in a scope matrix is whichever one the supply agreement drew. So the useful question on a real project is never what an AC block is in general — it is which equipment sits inside the line, and who owns the interface where it stops. Whether storage meets a co-located generator on the AC bus or shares a DC bus with it is a separate question, and the AC-coupled / DC-coupled entry's subject rather than this one's.
The conversion-and-transformer half of the block is a catalogue item, which makes its dimensions easy to check against published sheets. Power Electronics publishes its MV Skid Compact across 1,525-4,390 kVA at 40 C at 690 V LV, with MV winding options from 6.6 kV to 34.5 kV and LV terminals offered in nine steps from 480 V to 690 V — the top of that range is the 690 V reading of one current limit, and the same frame is published near 3,055 kVA ordered at 480 V; the twin-transformer version of the same family spans 3,050-8,780 kVA at 40 C.
EPC Power's turnkey skid station is quoted at 3-5.3 MVA in two-, three- and four-inverter variants, the transformer sized 3.0, 4.5 or 5.3 MVA to match, with MV terminals at 12.47, 13.8 or 34.5 kV. Its M Skid ladders ten 537 kVA inverters to 5,370 kVA at 690 VAC and 45 C behind a 4.3 or 5.3 MVA transformer.
SMA's MV power stations land in the same territory, roughly 2.3-4.1 MVA depending on LV voltage and ambient. A four-inverter EPC skid is quoted at about 38.5 by 8.5 by 9.25 feet and 48,000 lb, approximate and finalised per project — one crane pick, one foundation detail, repeated across the site.
How much battery rides behind one block is a duration decision rather than a property of the block. Take a 5 MVA block held at 0.95 power factor: about 4.75 MW, and four hours of that is roughly 19 MWh delivered at the AC terminals.
Work backwards through one-way conversion and the usable state-of-charge window and the DC nameplate required sits in the low twenties of MWh — four to five of today's roughly 5 MWh-class liquid-cooled enclosures, or a great many more of the 1-3 MWh air-cooled units earlier fleets were built from. Change the duration and the block's AC nameplate does not move at all; only the enclosure count does. That asymmetry is why power and energy end up procured, tested and guaranteed against different documents.
Why it matters in a real grid-scale project
Blocking is an engineering-cost decision before it is an electrical one. One block design carries one set of protection settings, one cable schedule, one foundation detail, one commissioning checklist and one factory acceptance procedure, and every megawatt after the first block reuses all of them.
It also parallelises the site: blocks are set, terminated and pre-commissioned independently, so schedule risk concentrates in the shared infrastructure — collection feeders, the main power transformer, the substation — instead of being smeared across every megawatt. And it makes the plant countable. Capital, spares, availability and outage exposure are budgeted per block and multiplied, which is why an owner's first question about an unfamiliar supplier is usually the block rating rather than the plant rating.
The second payoff is granularity of loss. Each block ties to the collection feeder through its own MV switching, so a fault inside one block is cleared by that block's device and the rest of the plant carries on — Power Electronics rates the switchgear in its compact station at 16 kA for 1 s, with 20 kA and 25 kA options, in a double-feeder arrangement that lets blocks sit along a loop rather than each needing a dedicated run back to the substation.
The exposure that arrangement buys is one block's worth: on a 100 MW plant assembled from about 5 MVA blocks, losing a block is on the order of 5% of AC capability, plus every megawatt-hour parked behind it, until it is repaired. Spares strategy follows the same unit — a spare transformer or a spare inverter module returns a block, and how fast it does so is most of what an availability guarantee is actually pricing.
Catalogue MV stations cluster around 1.5–5.3 MVA per transformer (PE MV Skid Compact 1,525–4,390 kVA at 40 °C and 690 V LV; EPC turnkey skid 3–5.3 MVA), and roughly four to five 5 MWh-class enclosures sit behind a ~5 MVA block at four-hour duration — change the duration and the count changes, not the block nameplate. The same product rates 1,415–4,075 kVA at 50 °C, so the rating only means something with its conditions attached. The efficiencies on the figure are EPC's own published numbers for its skid station — 98.4% inverter, CEC-weighted, and 99% transformer typical — and EPC notes overall efficiency varies with transformer design and operating point, so they are one vendor's product figures rather than properties of the stages. LV terminal voltage is 690 VAC on mainstream utility-scale blocks, 400–480 VAC in commercial and industrial designs, with the industry moving toward 720–850 VAC. Blocking buys one set of drawings, one factory test procedure, one spares list and one fault-isolation boundary; it does not buy a contract boundary. PE cites IEC 62271-212 / 62271-200 / 60076 / 61439-1 for the compact station in IEC markets, while in the US the grid-code duties are measured at the POI under IEEE 1547 for distribution and IEEE 2800 for transmission.
- Where the block ends
- At an AC terminal — enclosures, PCS and the LV/MV step-up transformer as one purchased unit; the POI sits further downstream
- Typical single-transformer block size
- Catalogue MV stations cluster ~1.5-5.3 MVA per transformer (PE MV Skid Compact 1,525-4,390 kVA @40 C at 690 V LV; EPC turnkey skid station 3-5.3 MVA); twin-transformer variants ~3,050-8,780 kVA @40 C
- Same product, warmer day
- PE publishes 1,525-4,390 kVA at 40 C at 690 V LV and 1,415-4,075 kVA at 50 C for the same compact station — the rating only means something with its conditions attached
- Terminal voltages offered
- LV commonly 480 V through 690 V; MV options 6.6-34.5 kV (EPC lists 12.47 / 13.8 / 34.5 kV). 34.5 kV / 690 V is a rated-voltage quotient of 50
- Enclosures behind one block
- Roughly four to five ~5 MWh-class enclosures for a ~5 MVA block at four-hour duration, once the usable window and one-way conversion are allowed for — duration changes the count, not the block nameplate
- Loss granularity
- A failed block on a 100 MW plant of ~5 MVA blocks is on the order of 5% of AC capability plus the energy behind it; PE rates the block's MV switchgear at 16 kA for 1 s (20 kA or 25 kA optional)
- Losses inside the block boundary
- EPC publishes 98.4% inverter (CEC-weighted) and 99% transformer typical for its skid station, noting overall efficiency varies with transformer design and operating point
- Assembly standards vs grid obligations
- PE cites IEC 62271-212 / 62271-200 / 60076 / 61439-1 for the compact station in IEC markets; in the US, grid-code duties are still measured at the POI under IEEE 1547 (distribution) and IEEE 2800 (transmission)
Typical values and standards
Every block nameplate quoted above is one row of a matrix rather than a constant. The same Power Electronics station published at 1,525-4,390 kVA at 40 C (690 V LV) is published at 1,415-4,075 kVA at 50 C, with a stated environmental envelope of -25 C to +50 C, derating above that, and 1,000 m maximum altitude; EPC's skid station is quoted at -20 C to +55 C (-40 C optional), IP54 and NEMA Type 3R equivalent, and elevations above 3,000 m.
The LV terminal voltage shifts the rating several times more than ambient does: on the same current limit, 690 V to 480 V costs about 30% (4,390 to 3,055 kVA) against roughly 7% between 40 C and 50 C. The reasons belong to the power conversion system's own entry. Read a block rating with its conditions attached and confirm those conditions are your site's, because the conditions travel with the number and the number does not travel without them.
Standards attach to the block as an assembly, not only to the devices inside it. In IEC markets Power Electronics lists its compact station against IEC 62271-212 for the prefabricated compact station itself, IEC 62271-200 for the MV switchgear, IEC 60076 for the transformer and IEC 61439-1 for the low-voltage assembly — four documents, and only one of them treats the block as a delivered object.
In North America the energy storage system is certified to UL 9540, while UL 9540A is the fire-propagation test method whose data feeds the installation rules, never a certification; which physical boundary was listed is the question an authority having jurisdiction will actually put.
None of these govern grid behaviour. In the United States those duties are measured at the point of interconnection under IEEE 1547 for distribution-connected resources and IEEE 2800 for transmission-connected inverter-based resources wherever they are adopted, and in every case under the applicable ISO/RTO and utility interconnection rules, which are what actually bind. No block-level certificate discharges any of them.
Two loss stages live inside the block boundary and both are usually published. EPC quotes 98.4% inverter efficiency (CEC-weighted) and 99% transformer efficiency typical for its skid station, with the caveat that overall efficiency varies with transformer design and inverter operating point. Whatever a block guarantees at its MV terminals is therefore already net of those two stages — and still gross of the collection feeders, the main power transformer and the site's auxiliary load.
How it shows up in specs, studies and contracts
Block-level guarantees are written at the block's own terminals, and a plant is not guaranteed by adding them up. The distance between the sum of the blocks and what the revenue meter registers is the medium-voltage collection system, the main power transformer and the auxiliary load, and someone has to own it — normally the integrator or the EPC contractor, through a wrap that converts block guarantees into a plant number.
Ask explicitly which boundary each figure is stated at: MVA at the block's MV terminals, MWh at the DC bus, or MW and MWh at the POI meter. Ask it of availability too. Block-hours available and MW-weighted plant availability are different metrics on the same site and can read several points apart, because one counts a failed block as one unit among N and the other counts the megawatts it took with it.
The capacity test does not respect the block boundary at all. It runs at the contracted meter, usually the POI, against contract energy under defined conditions, and it settles the whole plant in one measurement — so a weak block is not visible as a block and surfaces only as a shortfall against the plant number. Two consequences follow.
Per-block acceptance during commissioning is the only stage at which a bad block is cheap to find, so run those tests and keep the raw data. And check what the capacity warranty is baselined against, because a nameplate baseline hands an over-label COD test to the supplier as headroom: the start point later retests are compared to is then the label rather than what the plant actually measured on the day.
Common pitfalls
Augmentation is where the block boundary bites hardest, and the direction of the change decides how hard. Adding enclosures behind an existing block adds megawatt-hours without adding megawatts — the block's transformer, switchgear and cabling still see the same apparent power — so duration grows and the AC-side equipment is untouched, provided the DC bus, the conversion capacity and the enclosure count the block was engineered around can absorb it.
Adding blocks is the other move; it adds both, but it also consumes a collection-feeder position, a share of the main transformer and a piece of the interconnection limit, none of which come free once the plant is built. Which move a plant can still make in year eight is largely fixed on the day the block is specified, so write the intended augmentation path into the specification and make the supplier confirm in writing that the block will take it.
Two habits cause most of the avoidable trouble. The first is reading the block nameplate as deliverable power at your site, when the same product line publishes different figures for 40 C and 50 C and different figures again per LV terminal voltage. The second is assuming the block bought this year will still be purchasable in year seven.
An integrated AC-block product follows its vendor's roadmap — cell supplier, inverter platform and firmware can all move between orders — so a replacement block years in may be electrically compatible without being identical in controls or in spare parts. Where that matters, negotiate for the interface documentation and the settings alongside the hardware, and record which block revision sits at each position on site, because a long-lived plant generally ends up carrying more than one.
Building the plant in blocks means an outage only costs you the piece of equipment that failed.
In reality: The block is normally the smallest unit a plant loses, and much of what sits inside it is single-string — one step-up transformer, one MV switch, one set of skid controls — so a healthy battery fleet behind a failed transformer usually counts as unavailable along with it. On a 100 MW plant built from ~5 MVA blocks that is on the order of 5% of AC capability plus the energy parked behind that block until repair. Whether it reads as an availability deduction or as a capacity shortfall depends on which boundary the contract measures, and a plant-level capacity test will only ever show it as a shortfall against the plant number.
- Interactive: MV Skid Structure Interactive visual · bess.engineer
- Interactive: Energy Station Structure Interactive visual · bess.engineer
- DC block Glossary
- BESS commissioning and capacity testing Article
AC block, in context.
The Grid-Scale BESS course covers ac block — and the rest of the system — from the ground up, the way it actually gets deployed.