NC State and NYPA Ran a 1 MVA Solid-State Transformer on a Live 13.2 kV Feeder, But Not at Full Power

Key Facts
  • 1 MVA solid-state transformer run on a live utility feeder
  • 13.2 kV distribution feeder at EPRI's laboratory in Lenox, Massachusetts
  • 3 days of field testing in mid-June 2026
  • 10 energize and de-energize cycles at minimum
  • 98% peak efficiency, module level only, not grid to load
  • DE-EE0008450 is the DOE award funding the work

NC State University, the New York Power Authority and EPRI ran a 1 MVA solid-state transformer on a live 13.2 kV distribution feeder. The campaign lasted three days in mid-June 2026. The test took place at EPRI’s power delivery laboratory in Lenox, Massachusetts. The unit fed an electric vehicle load, pushed reactive power back onto the feeder, and survived at least 10 energize and de-energize cycles.

“This is the first independently verified, megawatt-class solid-state transformer validated on a live utility distribution feeder,” said Srdjan Lukic, the Lampe Distinguished Professor of Electrical and Computer Engineering at NC State and the project’s principal investigator. The Department of Energy’s Transportation Technologies Office funded the work under award DE-EE0008450. NYPA paid for construction and field testing.

The claim is narrow on purpose, and the qualifier matters as much as the result. The unit did not run at full power. Site load constraints capped what the team could draw.

What the Demonstration Actually Proved

A solid-state transformer folds two jobs into one box. A conventional installation steps voltage down with a magnetic transformer, then converts AC to DC with a separate rectifier. An SST does both with power electronics and high-frequency isolation, and it does them in both directions. This unit uses 15 kV silicon-carbide MOSFETs to reach medium voltage without a cascade of lower-voltage stages.

The Lenox campaign was a connection test before it was a performance test. “The key goal of this demonstration was simply to see: Can we connect this system to a distribution feed?” Lukic said. The answer was yes, repeatedly, and without drama. “Connecting the device, and you don’t need see sparks. That’s a great achievement,” said Ramadan Elmoudi, senior research, technology and development engineer at NYPA and the utility’s project lead.

Two results go beyond simply staying online. The unit delivered real power to a vehicle and reactive power to the grid in the same operation. That bidirectional behavior is what makes an SST worth its complexity. Lukic also reported low harmonic distortion. That claim matters. Power electronics at distribution voltage usually create harmonics rather than cure them.

Why It Matters

The pull for this technology is not efficiency. It is siting. A 400 kW vehicle charger or a data center hall behind a conventional feed needs a transformer, a rectifier, switchgear and the pad to hold all of it. An SST collapses that footprint. It also makes voltage regulation and bidirectional flow properties of the same device. NYPA therefore calls it a tool in its clean energy toolbox, not a replacement part.

The research line is long. NC State’s FREEDM Center started with a 10 kVA laboratory prototype and worked up through units above 20 kVA before this one. Neither NC State nor NYPA plans to sell the device. Their stated aim is to move vendors and utilities off the fence. The Lenox result is the first independent evidence they can hand over.

Utilities buying transformers today are choosing among products with decades of failure data behind them. The honest reading of Lenox is that a megawatt-class SST now clears the first gate. The gates that decide whether one ever sits on a real feeder for real money have not been opened yet.

What It Did Not Prove

The efficiency number circulating with this story needs care. NC State’s own August release contrasts roughly 96 percent for a conventional transformer-and-rectifier chain against roughly 98 percent for the SST. Lukic’s account is narrower. “On the actual demonstration site, we didn’t focus on the efficiency of the unit from the grid to the load,” he said. The 98 percent figure describes a single module converting medium voltage to the DC bus. That module efficiency “peaks out at about 98%,” he said. Nobody measured grid-to-load efficiency for the full 1 MVA system.

Nor did three days say anything about the thirty years a utility expects from a transformer. “We have to go from where we are, which is demonstrating that the system operates in the target environment, to a device where the end user has confidence that these units will last as long as a traditional transformer,” Lukic said.

Elmoudi named the gap that blocks procurement rather than engineering. “I believe the next big step is going to be standardization because we don’t have any standards for SST,” he said. He then listed the questions no standard yet answers. “If we have a lightning, what’s going to happen to the SST? If we have through fault, what’s going to happen to the SST?”

The qualification list is long and mostly untouched: overvoltage and undervoltage tolerance, short-circuit and overload protection, lightning and switching surge response, harmonic distortion under load, thermal cycling, accelerated aging, and cybersecurity. The custom isolation stage still needs its own round of development.

Critical Perspective

The headline claim is real and also narrow. Three days on a feeder, below full power, answers one question: will the device connect and stay connected. It does not answer whether a solid-state transformer survives a decade of thermal cycling, a lightning strike, or a through fault. Elmoudi asked those questions himself, which is a good sign about the team and a bad sign about the technology’s readiness.

The efficiency number deserves more scrutiny than it is getting. NC State’s release sets roughly 96 percent for a conventional chain against roughly 98 percent for the SST. Lukic’s own account undercuts that comparison. The 98 percent is a module peak, grid-to-load efficiency was not measured at Lenox, and a peak is not an operating average. Anyone modeling lifetime losses from the 98 percent figure is using a number the demonstration did not produce.

The gap that decides this technology is not technical. Elmoudi named standardization as the next step, and he is right that no standards body has written an SST specification. A utility cannot buy against a spec that does not exist, cannot rate-base an asset with no service history, and cannot ask a lineman to work near a cabinet whose failure modes are undocumented. The qualification list in this account runs to a dozen test regimes, and none of them has been run.

One number is missing from every account of this demonstration: cost. A distribution transformer is a commodity with a century of manufacturing behind it. This device replaces it with 15 kV silicon-carbide MOSFETs, custom isolation and a control system. Neither NC State nor NYPA plans to sell it, so the party that has to justify the price premium is a vendor who has not yet been named.

Sources

Related Coverage

Research Implications
Why it matters

DE-EE0008450 is the DOE award funding the work

Related post