How EDF Normalized Grid Data Across 50 Sites and 16 GW

EDF Power Solutions North America manages 16 gigawatts of power across wind, solar, and battery energy storage systems, with equipment from dozens of OEMs that was never designed to interoperate. To bring all of that data into a common format, the company deployed an edge solution across more than 50 sites and adopted the IEC 61850 standard. During a panel at IIoT World Energy Day 2026, Brenna Wood of EDF described what that effort involved, what it cost, and what other operators considering the same path should expect. Andrew Foster of IOTech Systems described how standards like MQTT Sparkplug and OPC UA fit into the normalization architecture.

1. Why Is Grid Data So Hard to Normalize?

Grid data is hard to normalize because equipment from different OEMs uses different protocols, different naming conventions, and different data models, and a 50-millisecond timestamp drift between two sensors is enough to compromise AI outputs.

Some systems speak Modbus. Others speak DNP3, or OPC. Many are proprietary. But the protocol layer is only part of it. Naming conventions vary across equipment that was never designed to work together. The same physical measurement, coming from two different vendors’ equipment, may carry completely different labels. Before any of this data can be used efficiently, it needs to be normalized and contextualized.

The context problem goes deeper than labels. A 50-millisecond drift between two sensors is a significant problem when running AI. Ontology matters too: how all data maps into a semantic layer, a common language for what an asset is, how it behaves, and what its relationships are. And then there is the edge-cloud boundary, the question of what gets processed locally versus what moves to the cloud.

2. What Did EDF’s Normalization Across 50 Sites Involve?

EDF deployed IEC 61850 across more than 50 sites to create a common data format for its 16 GW fleet, but the technical standard required business-level naming conventions alongside it before the first data point could flow.

Before the normalization project, people across the organization were developing their own analytics, with significant duplication of effort. When fleet engineers encountered a new problem, if the relevant data had not been tagged up front, it meant going back to normalize another new data set each time.

IEC 61850 provided the technical framework. But a technical standard alone required business-level naming conventions alongside it, understandable by control room operators, fleet engineers, and site personnel. There were many factors to weigh before the first data point could flow, including the trade-offs between upfront costs and ongoing run costs.

One point that came out of the discussion: data standards should be part of the procurement process. Including compliance with IEC 61850 or whatever standard the operator chooses in the bidding process makes it easier for vendors from day one. Considering this during procurement rather than after installation saves significant effort.

3. How Do IEC 61850, MQTT Sparkplug, and OPC UA Work Together?

IOTech Systems takes data from heterogeneous equipment, normalizes it into MQTT Sparkplug for distribution across system components, and remaps it into OPC UA to provide an interface for SCADA systems and other applications.

Andrew Foster of IOTech Systems described an approach where data coming off heterogeneous equipment is normalized into formats like MQTT Sparkplug for distribution across components within the system. From there, it can be remapped into OPC UA to provide an interface for SCADA systems, UIs, and other components.

“Standards support plug-and-play interoperability,” Foster said. They prevent vendor lock-in and make “the integration phases of the project a lot easier than proprietary APIs.” For systems designed to operate 20 to 25 years, that long-term flexibility matters.

4. Why Does Grid Data Normalization Continue After Deployment?

Grid data normalization continues after deployment because business needs evolve, operating models change, and fleet engineers ask questions about problems nobody anticipated when the system was first configured.

At EDF, the company’s role in operating battery assets changed mid-course, requiring the team to take on the balancing of the battery. That shift demanded a whole different depth of data. What had been configured for one operating model did not cover the new one.

The standard provides the framework. But the mapping of business context onto that framework is a continuous process. Brenna Wood described how new questions from fleet engineers about unanticipated problems require ongoing normalization work beyond the initial deployment.

Standards-Based vs. Proprietary Approach to Grid Data

 This article is based on the IIoT World Energy Day 2026 panel “Turning Industrial Data into Energy Insight: Scalable Edge Solutions for Modern Grids,” featuring Andrew Foster (IOTech Systems), Brenna Wood (EDF Power Solutions North America), Cody Falcon (ABB), and Janko Isidorovic (Fluence Energy), with moderation by Hamish Mackenzie (IIoT World). AI tools were used to help summarize and organize the content. Reviewed and edited by the IIoT World editorial team.

Sponsored by IOTech Systems.

Watch the full panel discussion