Trusted Data Foundations Power AI Governance
Utilities across the globe are under increasing pressure to modernize their grids, integrate renewable energy sources, and maintain high standards of reliability and resilience. However, these ambitions are often challenged by fragmented data landscapes, legacy systems, and disconnected IT and OT environments. The DNV-UAI Roundtable “From Silos to Synergy: Building a Trusted Data Foundation for AI Ready Utilities” convened experts from DNV, Avista, and the Utility Analytics Institute to explore how utilities can overcome these barriers and build a trusted data foundation that supports robust AI governance, enabling utilities to meet the demands of artificial intelligence and advanced analytics.
AI Governance Enables Strategic Solutions
Utilities are driven by several strategic goals: modernizing the grid to enhance reliability and flexibility, integrating renewables to support sustainability targets, improving operational efficiency, and ensuring compliance with evolving regulatory requirements. Yet, the path to achieving these goals is fraught with challenges. Legacy systems and siloed platforms impede data sharing and interoperability, while data fragmentation complicates analytics and informed decision-making. The sector also faces a skills gap, as new technologies demand upskilling and robust change management. Vendor lock-in remains a persistent issue, limiting flexibility and increasing costs, and the proliferation of digital streams introduces new cybersecurity vulnerabilities.
Despite these obstacles, there are significant opportunities for utilities willing to embrace change. Artificial intelligence and machine learning are enabling predictive maintenance, outage prevention, and automation of customer service, while the adoption of data standards such as the Common Information Model (CIM) is fostering interoperability and reducing vendor dependency. Cloud migration offers scalability and access to advanced analytics, further empowering utilities to innovate and strengthen their AI governance framework.
We are at a major breakthrough in regard to AI technology. Some claim that this is as big as the Industrial Revolution. I think it’s going to be changing and impacting everything that we do, including data management moving forward.
Hossein Nikdel, Avista
Modular Architecture Enhances AI Governance
A standardized reference framework is essential to support real-time grid operations, planning, asset management, and digitization across transmission and distribution (T/D) system operators (T/DSOs) and generation entities. It enables establishing an internal AI governance framework, defining new processes, data requirements, and accountability structures that align with enterprise-wide digital transformation goals.
A strong AI governance framework includes building a secure, modular architecture and achieving effective IT/OT integration, which are foundational to successful grid modernization. Utilities should establish clear segregation between IT and OT networks, leveraging demilitarized zones (DMZs) to ensure controlled and secure communication across domains. In the below reference architecture, IT networks host systems such as Geographic Information Systems (GIS) and Enterprise Asset Management (EAM), serving a broad user base and utilizing integration layers for efficient data exchange. OT networks manage mission-critical platforms like Advanced Distribution Management Systems (ADMS) and operate with restricted access for select users. Additional networks, such as telecoms, support both on-premise and cloud solutions, with DMZs acting as secure buffers between environments.
To support IT/OT convergence, it is essential to develop integration diagrams that visualize data flows and use cases, thereby reducing inefficiencies and enhancing data accuracy. Grouping integrations into functional flows simplifies system complexity, while defining integration methods and communication protocols such as REST APIs for command-based interactions and distributed event streaming data platforms ensures robust, real-time operations. Modular design principles further enable utilities to avoid vendor lock-in and facilitate the seamless addition of new services as business needs evolve.
Modular architecture enables seamless data exchange between systems, allowing organizations to reuse data where needed rather than duplicating the same information across multiple platforms. By centralizing and standardizing data flows, utilities can reduce redundancy, improve data quality, and support more efficient operations and analytics.
Marellie Akoury-Shima, DNV
Cybersecurity must be embedded by design, including conducting comprehensive risk assessments and aligning with international standards such as ISO 27001, IEC 62443, IEC 62351, and the NIST Cybersecurity Framework (International Organization for Standardization & International Electrotechnical Commission, 2022; International Electrotechnical Commission, n.d.; International Electrotechnical Commission, n.d.). Key measures include implementing access controls, multi-factor authentication, and data encryption (both at rest and in transit), as well as network segmentation, firewalls, and data loss prevention mechanisms. Utilities should also deploy intrusion detection systems, log monitoring, and incident response plans, alongside cryptographic hashes and digital signatures to ensure system integrity. Finally, load balancing, redundancy, and disaster recovery planning are vital for maintaining high resource availability and operational resilience.
Data Standards Enable Scalable AI Governance
A deep dive into data standards and analytics reveals that establishing a single source of truth is foundational for AI readiness. The adoption of data standards enables consistent data quality and interoperability across systems, while robust AI governance, which encompasses data governance, ensures the implementation of polices across the organization such as standardized ontologies and object naming conventions. Automated CIM testing tools, essential for validating vendor implementations and business processes, are used to ensure that data models align with enterprise schemas and regulatory requirements.
You must have the right structures and the quality of data. So, if you have that incorrect, then your AI will not work the way you want it.
Bas Kruimer, DNV
CIM Standardization Builds a Foundation for AI Governance
A strong foundation for artificial intelligence in utilities relies on having a structured, standardized approach to organizing and interpreting data across diverse systems. By normalizing data and enabling seamless integration between platforms, organizations can accelerate analytics, ensure data quality, and support scalability and reusability. This approach allows for advanced applications such as grid optimization and predictive maintenance, while also making it easier to validate information and adapt to frequent data updates. Ontologies grounded in robust data standards are essential for reasoning and interoperability, as they provide the structure needed to enable inference and interoperability across systems. While AI cannot replace the human expertise required to build effective ontologies, it can play a valuable role in supporting their creation, alignment, and ongoing maintenance especially as datasets grow larger and more complex. Ultimately, establishing consistent data standards and robust integration workflows within an AI governance framework empowers utilities to unlock the full potential of AI, driving operational efficiency and innovation.
Operationalizing a Trusted Data Foundation for AI Readiness
Real-world implementation experiences, such as those shared by Avista, highlight the practical steps and lessons learned in building a trusted data foundation. Avista’s Grid Model Manager (GMM), currently under development, will serve as a centralized repository for grid models, supporting quality assurance and downstream use cases. The development of a CIM model export, in collaboration with internal and external experts, has enabled plug-and-play functionality for applications and reduced integration complexity. Change management, internal buy-in, and capacity building have been critical to success, with expert support in extending standards and validating models. Implementation and testing typically require one to two years, but subsequent projects benefit from established tools and organizational expertise, accelerating future deployments.
We are fundamentally transforming the way we work by establishing a platform for true cross-domain collaboration. Rather than relying on a patchwork of proprietary languages and schemas, we are embedding the Common Information Model (CIM) at the core of our processes. This approach enables a single, standardized language, dramatically reducing integration complexity and fostering interoperability. By doing so, we create a single, trusted source of truth for our data—empowering teams across the organization to make informed decisions, accelerate innovation, and build a resilient foundation for future digital initiatives.
Erik Lee, Avista
Strategic Insights for AI Ready Utilities
Key takeaways from the roundtable and utility industry practice emphasize that a robust data foundation is critical for successful AI and analytics initiatives. Modular, secure architectures provide the flexibility, scalability, and resilience needed for modern utility operations. The adoption of data standards such as CIM builds a foundation for AI governance, drives interoperability, and reduces vendor lock-in, while effective change management ensures that organizations are equipped to navigate technological transitions. Continuous collaboration with peers, vendors, and regulators is essential to advancing best practices and fostering innovation. As agentic AI and large language models (LLMs) reshape utility operations, new approaches to data integration and event synchronization will be required. By embracing these principles, utilities can move from silos to synergy, building trusted data foundations to become AI ready.
Generative AI was utilized to summarize the roundtable transcript.
References
DNV, (2025, September 16). DNV-UAI MEMBER-ONLY ROUNDTABLE From Silos to Synergy: Building a Trusted Data Foundation for AI-Ready Utilities. https://utilityanalytics.com/document/from-silos-to-synergy-building-a-trusted-data-foundation-for-ai-ready-utilities/ (member-only access)
International Organization for Standardization & International Technical Commission. (2022). Information security, cybersecurity and privacy protection – Information security management systems – Requirements. (ISO/IEC Standard No. 27001:2022). https://www.iso.org/standard/27001
International Electrotechnical Commission. (n.d.). IEC 62443 Series of Standards. Security for industrial automation and control systems. https://www.isa.org/standards-and-publications/isa-standards/isa-iec-62443-series-of-standards
International Electrotechnical Commission. (n.d.). IEC 62351 series: Power systems management and associated information exchange – Data and communications security. https://webstore.iec.ch/en/products/?p=1&f=eyJkYXRlUmFuZ2VzIjp7fSwidGVybXMiOnsiaGVhZGVyIjp7IjEiOlsiSUVDIl19fSwidmFsaWRPbmx5Ijp0cnVlLCJwdWJsaWNhdGlvbklkcyI6bnVsbCwic2hvd1RyZiI6dHJ1ZSwiZGlzcGxheU1vZGUiOiJncmlkIn0=
National Institute of Standards and Technology. (n.d.). Cybersecurity Framework: RMA Conference. Computer Security Resource Center. https://csrc.nist.gov/Projects/cybersecurity-framework
About the Authors
Hossein Nikdel
Hossein Nikdel is the Director of Digital Innovation for Utility Intelligence and Chief Enterprise Architect at Avista. He is a Ph.D. candidate in Applied Mathematics and Computer Science, with over 37 years of experience spanning industries such as commercial software development, services, manufacturing, government, and energy management.
Erik Lee
Erik Lee, Principal Distribution Planning Modeling & Tools Engineer at Avista, received his BSEE at Washington State University in 2006 and has been working with electric distribution models most of his 18-year career. He holds a Professional Engineer’s license in WA state and has been involved in various projects such as standing up Avista’s current Distribution Management System (DMS), creating and maintaining the Synergi Electric model building process, leading the company’s foray into IEC Common Information Model (CIM) modeling, working on Advanced Metering Infrastructure (AMI) integration with Outage Management System (OMS) and our Aveva PI system, and using C# to develop various tools and processes to help address various engineering, operations & planning needs such as load forecasting and temporal load characterization.
Kristina Kelly
Kristina Kelly, Director for Energy Insights in Medford, Massachusetts, has been working at DNV since June 2008 and has fifteen years of years of experience in energy consulting and twelve in designing and managing projects that examine energy and demand savings potential and how market structures and program interventions influence program results. Ms. Kelly leads DNV’s Non-Wires Alternatives service, overseeing technical planning and design projects for utilities across North America and research into the costs and benefits of non-wires solutions and portfolios. In Massachusetts she provides oversight as the Markets research area lead ensuring market structure and resource adoption research efforts provide the Massachusetts Program Administrators and other stakeholders with valuable market insights. In addition to project and contract management, her areas of expertise include presenting and disseminating information from a variety of sources to support market characterization, cost benefit analysis, and market potential, and energy and demand savings modeling from Distributed Energy Resources and Demand Side Management Strategies and measures. Ms. Kelly holds a Master’s in Economics from Boston University.
Marellie Akoury-Shima CEng MIET, TOGAF®
Marellie Akoury-Shima is a Principal Consultant in DNV’s Energy Strategy Advisory team, based in London, UK. A Chartered Engineer with over 18 years of experience in the energy sector, she brings deep expertise in delivering strategic and technical guidance to support the global energy transition.
Marellie serves as the Global Practice Lead for Digital System Operation, where she leads a worldwide network of experts in shaping DNV’s global strategy and consulting services. Her work spans multiple regions, aligning cross-functional teams and client priorities to drive innovation, operational excellence, and measurable impact across the energy value chain.
At DNV, her primary focus is on the Energy Transition, where she leads initiatives in strategy development, system operation, technical specification, and project implementation of Operational Technology (OT) systems. Her work emphasizes the integration of OT with IT systems, with a strong focus on cybersecurity, IT/OT convergence, and data management. She also advises system operators on developing strategic roadmaps to achieve Net Zero targets.
Marellie also plays a key role as Business Lead in DNV’s AI Champions network, where she helps shape the company’s approach to artificial intelligence, fostering innovation and responsible adoption of AI technologies across business units.
Before joining DNV, Marellie held several key roles at National Grid’s Electricity National Control Centre in the UK, where she worked on regulation, energy markets, advanced grid operations, operational tools, and visualization technologies.
In addition to her consulting work, Marellie is an active contributor to international standards development. She participates in IEC TC 57 Working Groups 13, 14, 16, and 19, supporting the evolution of the Common Information Model (CIM) standard. She also serves on the British Standards Institute’s CIM Governance Group and has authored numerous papers on the subject.
Chuck Juhasz
Chuck Juhasz, Senior Principal Consultant for DNV, has 25 years of professional development experience and is an enterprise system architect and software developer with over two decades of experience delivering mission-critical enterprise systems across a variety of industries and technologies. He has worked on “Big Data” Smart Grid products and on the implementation and integration of large-scale enterprise system integrations. Chuck has been involved in all aspects of the software product lifecycle. He has integrated distributed systems utilizing a variety of technologies including Enterprise Services Buses, Web Services, various IP protocols, and file-based EAI. Chuck has also helped to develop IoT enterprise and consumer products from hardware to software integration and mobile app development.
Bas Kruimer
Bas Kruimer is a seasoned expert in Digital Grid Operations and Cyber Security, formerly serving as Business Director at DNV Energy Systems in The Netherlands. His focus has included digital systems, automation strategy and deployment, with special attention to utility cyber security. With a background in SCADA, SA/IEC61850, Smart Metering, and OT Cyber Security, Bas has supported utilities in their diverse Next Generation Grid Operations, network control, grid automation, digital transformation, and cyber security challenges.
Graduated as power engineer from Delft University of Technology end 80’s Bas started his international career in electrical power transmission and distribution 35 years ago at ABB Network Control & Protection, followed by KEMA/KEMA Quality, Eneco/Joulz, Quanta Technology and Accenture/Accenture Security.
In February 2023 Bas was appointed Fellow of Delft University of Technology in the department of Electrical Sustainable Energy – Intelligent Electrical Power Grids with Prof. Peter Palensky and Dr. Alex Stefanov on the topic Control Center of the Future & Grid Cyber Resilience supporting Master & PhD students in their research projects, connecting to industry and setting up partnerships. Bas is also a regular speaker, conference chair and trainer on Next Generation Grid Operations & Cyber Security.
