One Customer's Guide to Data Modeling with Cognite Data Fusion®

Skagerak Kraft, a leading hydropower producer with 52 plants generating 5.9 TWh annually, faced significant challenges in managing its growing operational data volumes.

With data scattered across systems, the company needed help to extract meaningful insights from critical datasets like reservoir water levels, sensor readings, and operational logs. Additionally, end users found it challenging to navigate overlapping and redundant sensor data, which complicated efforts to make informed decisions about equipment performance, water management, and energy trading.

Skagerak recognized that solving these issues required a structured approach to data integration and contextualization.

The goal was to create a unified framework that met industry standards and aligned with operational realities, empowering employees to access and trust the data they needed.

Adopting a DataOps Mindset: A Foundation for Success

To address its data challenges, Skagerak Kraft recognized the need for a fundamental mindset shift—moving from a system-centric to a data-centric approach. This evolution required fostering a DataOps culture that emphasized collaboration, governance, and the strategic management of data as a product. The journey was both a top-down directive from leadership and a grassroots effort driven by domain experts.

Building Cross-Functional Collaboration

Skagerak began by recruiting a diverse group of participants spanning all business domains. For nearly a year, this team dedicated a few hours each week to developing a shared understanding of data. Starting with simple tools like Excel for data cataloging and whiteboards for sketching information models, these sessions evolved into sophisticated discussions around data governance, modeling, and operations.

The collaborative effort bridged organizational silos, with participants bringing insights back to their respective domains and leadership receiving continuous updates. This approach fostered alignment across the company, ensuring that strategic goals and operational realities informed the project.

Establishing Core Principles

The team defined two guiding principles for managing data effectively:

These principles informed the operationalization of DataOps at Skagerak, with roles and responsibilities clearly outlined. A new data owner role was introduced, responsible for identifying data management needs and overseeing the creation of data products. By focusing on data as a product rather than an IT system add-on, Skagerak ensured that data could be processed, shared, and utilized effectively across domains.

Choosing Cognite Data Fusion to Scale DataOps

As the scope of Skagerak’s data initiatives grew, it became clear that the existing tools and team structure couldn’t meet the organization’s ambitions. Instead of building a custom solution, Skagerak adopted Cognite Data Fusion® because it provided domain experts with the scalability and quality needed to create and manage data products. It allowed Skagerak to increase the number of simultaneous data initiatives without hiring additional developers, ultimately delivering a better and more cost-effective solution than could have been built in-house.

Empowering End Users with Data Access

Skagerak transformed the way engineers accessed and used data:

• Initially, data was only available directly from SCADA systems.
• Next, mobile devices enabled easier access to raw data in the field.
• Eventually, preprocessed, enriched, and verified data was introduced as a product to ensure reliability.
• Finally, Skagerak integrated these data products into Cognite Data Fusion®, providing users with a comprehensive suite of tools and analysis capabilities.

This evolution empowered end users to focus on extracting value from data rather than struggling with its accessibility or quality.

Using Cognite Data Fusion to Enable Scalable and Effective Data Modeling

By leveraging Cognite Data Fusion’s robust data modeling capabilities and tools, Skagerak built a flexible, scalable data infrastructure tailored to its operational needs. Here’s how they achieved this:

Centralized Data Management with Cognite Data Fusion®

Cognite Data Fusion® became the central hub for Skagerak’s data, allowing the data team to make information about major assets and time series easily accessible. Data consumers within Skagerak accessed this data directly through Cognite’s industrial tools, such as Cognite Charts, or third-party tools. This centralized approach ensured all stakeholders had consistent access to accurate, actionable data.

A Layered Data Modeling Approach

Skagerak adopted a layered data modeling framework, which includes:

1. Enterprise data model: Drawing inspiration from standards like the Common Information Model (CIM), Skagerak created an enterprise data model, spanning several business domains, describing the most central data objects to meet operational needs. For instance, the model integrated metadata, relationships, and groupings unique to the hydropower industry.
2. Solution Models: Specialized models are being purpose-built for specific use cases like power trading or asset intelligence. Solution models combine elements from the enterprise model and enrich them with additional data tailored to specific operational needs, such as predicting equipment lifespan or calculating reservoir water levels.

A Practical Data Modeling Lifecycle

Skagerak developed its data models using an iterative, four-step lifecycle:

1. Conceptualization: Domain experts collaborated to scope and define the core components of the enterprise data model, focusing on key structural data such as power plants and reservoirs. Lively discussions, such as defining a "watercourse," underscored the importance of this foundational work.
2. Definition and Implementation: Skagerak leveraged NEAT, an open-source tool developed by Cognite, to translate conceptual models into technical formats compatible with Cognite Data Fusion®. Designed for domain experts, information architects, and developers, NEAT simplifies the creation, enrichment, transformation, and onboarding of data models. By using Excel as a front-end interface, NEAT empowered Skagerak’s domain experts to actively contribute to data model development without needing coding expertise.
3. Population and Deployment: Once defined, Skagerak began to populate the models with data and deploy them into Cognite Data Fusion®. This will enable Skagerak to transition from asset-centric to model-centric data operations, ensuring scalability and consistency.
4. Iteration and Refinement: Skagerak will look to continuously improve its data models by iterating based on user feedback and operational needs. NEAT allows for seamless updates, enabling domain experts and information architects to expand and refine the models easily.

Impact: Transforming Data into a Strategic Asset

This case highlights how effective data modeling is critical for companies seeking to scale their digital transformation while delivering tangible value to end users.

The enterprise data model allowed Skagerak to simplify and standardize complex data scenarios, such as managing multiple sensors measuring the same reservoir’s water level. By iterating over versions of this data (e.g., using a median of sensor readings or incorporating physical models), Skagerak delivered increasingly precise outputs without requiring users to adapt their workflows.

Additionally, Skagerak began developing solution models to enhance operational capabilities further. For example, asset intelligence models tracked specific metrics, predicted equipment lifespans, and linked raw time series data to actionable insights.

A key outcome of this initiative was the upskilling of Skagerak’s domain experts. Through tools like NEAT and collaborative modeling sessions, these experts became active participants in the data modeling process. This improved the quality of the models and fostered a culture of data ownership and innovation.

With Cognite’s data modeling capabilities, Skagerak Kraft achieved significant improvements in its data management and operational performance:

• Improved Data Navigation: The enterprise data model streamlined sensor data, guiding users to the correct time series for reservoir levels and other key measurements.
• Enhanced Operational Insights: Models enabled real-time asset performance monitoring, reducing wear and tear on critical equipment and optimizing water flow for energy production.
• Accelerated Digital Transformation: By establishing an effective DataOps framework, Skagerak empowered employees to access high-quality data and deliver advanced use cases faster.

Skagerak’s decision to adopt Cognite Data Fusion was pivotal. Cognite Data Fusion’s ability to support multiple data models, enable real-time data population, and integrate with advanced tools allowed Skagerak to scale its data initiatives without hiring additional developers. The platform’s robustness and ease of use ensured that Skagerak could meet its operational goals more effectively and at a lower cost than building an in-house solution.

By leveraging Cognite Data Fusion’s powerful data modeling capabilities, Skagerak transformed its fragmented data into a unified, actionable resource. This enabled it to streamline workflows, empower users, and drive digital transformation across its operations.