Hook: Vendor SST decks keep promising denser AI halls and smarter feeders. Mid-June 2026, a different claim landed: NC State + NYPA + EPRI ran a university-built, roughly 1 MW / 1-MVA-class solid-state transformer on a live 13.2 kV utility distribution feeder at EPRI’s Lenox, MA power-delivery lab—and called it the first independently verified megawatt-class SST validated under real distribution conditions (POWER Magazine, Sept 8, 2026; NC State news, Aug 18, 2026). This is field-interface proof—not another product launch.
Key takeaway: Over a three-day campaign, the prototype was energized/de-energized ≥10 times, supplied real power to an EV load, and injected reactive power on operator command with claimed high power quality / low harmonic distortion. Lenox was not a full-power efficiency run—site load limits capped what they could draw. The often-quoted ~98% peak is a lab module figure (MV connection → SST DC bus), not measured grid-to-load for the full 1-MVA unit at Lenox.
What happened
Partners split cleanly: NC State designed and built the power-electronics hardware (PI Srdjan Lukic); NYPA funded construction and field testing (lead Ramadan Elmoudi); EPRI supervised the demo at Lenox. DOE Transportation Technologies Office award DE-EE0008450 seeded the path from lab prototype to feeder-capable hardware; NYPA paid to get it built and tested in the field.
- Site: EPRI Lenox, MA — 15-kV-class line operating at 13.2 kV L-L; mid-June 2026, three days.
- Tests: real power to an EV; reactive injection under operator command; repeated connect/disconnect without “sparks” drama (Elmoudi’s utility bar for first acceptance).
- Students named in the NC State release: Oscar Montes, Nazmul Hassan, David Dadzie—hardware built largely by graduate students, not a vendor skunkworks.
- No self-commercialization: NC State/NYPA intend EPRI to help share results with utilities and vendors so someone else can productize.
Engineering mechanism: why “live feeder” is the hard gate
Connecting PE to a medium-voltage feeder is a different problem than another 480 V lab brick. One device cannot block full feeder voltage; coordination and isolation decide whether the grid accepts you.
ol style="margin:0 0 1.2em;padding-left:1.4em;color:#E8EEF4;">NC State series-stacks smaller PE modules on the MV side so the 13.2 kV stress is shared across devices rather than asking one switch to hold the whole feeder. That is the architecture move that turns FREEDM-scale lab SST work into something that can even attempt a live distribution interface.
Series stacking creates a second problem: coordinate modules and isolate MV from LV/DC. Lukic called the team’s custom isolation transformer “one of the key innovations” and “really that last hurdle” solved with NYPA’s help—the stage that still needs deeper qualification before utilities treat it as decades-class gear.
- 3) Grid-interface functions, not just step-down
Because the converter sits at the feeder, the same stack can deliver real power to a load and inject reactive power on command—functions conventional transformers leave to separate devices. Lenox exercised both paths and reported low harmonic distortion on injected current. - 4) Efficiency: read the fine print before you model loss
Lukic told POWER they did not focus on grid-to-load efficiency at Lenox—instrument resolution and site priorities favored “can we connect and deliver.” Lab characterization of an individual module from MV connection to the SST DC bus “peaks out at about 98%.” That is not a measured full-system grid-to-load number for the 1-MVA unit on the feeder. NC State’s press contrast (~96% conventional transformer+rectifier vs ~98% SST / ~2% loss) is a useful order-of-magnitude framing—treat it carefully against the POWER interview’s module-level precision (mgrid.org nuance, Sept 9, 2026).
Misconception: “Another SST product launch—same beat as DG Matrix, Enphase IQ SST, Infineon×Eaton MVSST, or ABB Infinitus.”
Reality check: Those pieces sell vendor density, bricks, SiC-inside-OEM platforms, or source-to-rack block chains. This story is an independently supervised field validation of a university megawatt-class SST on a live utility feeder. No SKU, no RFQ sheet, no mass-production date—proof that the architecture can connect, cycle, and support real/reactive power at 13.2 kV without the grid rejecting it.
Limits / 待核:
- “First independently verified…” — partner claim (Lukic via NC State/POWER); no third-party census of all prior MW-class feeder demos reviewed here.
- Not full power at Lenox — site load limits; campaign prioritized connection and delivery over rated-power soak.
- ~98% peak — module-level lab path (MV→SST DC bus), not full 1-MVA grid-to-load at Lenox; NC State ~96% vs ~98% press contrast is order-of-magnitude, not a Lenox meter reading.
- Standards gap — Elmoudi: no SST standards yet for design/build/test/commission/maintain (IEEE or peer still needed).
- Survivability open — lightning, through-fault, isolation-stage qualification, long-term thermal/aging; EPRI Dec 2025 white paper still framed SST as largely pilot/prototype (per POWER).
- Utility comfort — decades of magnetic transformer + switchgear culture vs PE at the distribution interface; three days ≠ thirty years of failure data.
Takeaway judgment
Markets discussed—data centers (Lukic: possible killer app), EV charging, renewables/storage, microgrids, compact urban footprint—still need a vendor, a standard, and a survivability dossier. Lenox answers a narrower question that vendors cannot self-certify into existence:
Can a megawatt-class, series-stacked SST with custom MV isolation connect to a live 13.2 kV feeder, cycle repeatedly, and deliver real plus reactive power without the grid rejecting it?
Watch three checkpoints before you treat MW-class SSTs as procurement-ready:
ol style="margin:0 0 1.2em;padding-left:1.4em;color:#E8EEF4;">Related reading on this beat: SST / 800 VDC architecture explainers on The Engineering Core / @TheEngineeringCore-v; Traditional Chinese hardcore versions on vkinngworld and @Vkinng.
Sources:
- POWER Magazine — Sonal C. Patel, “Major Leap for Solid-State Transformers: Megawatt-Class Unit Demonstrated on Live Feeder” (Sept 8, 2026)
- NC State University — “NC State, New York Power Authority and EPRI Demonstrate Solid State Transformer Success Under Real World Conditions” (Aug 18, 2026)
- mgrid.org — “NC State and NYPA Ran a 1 MVA Solid-State Transformer on a Live 13.2 kV Feeder, But Not at Full Power” (Sept 9, 2026)
