Visualising Big Data for Energy Infrastructure
Designing a dashboard interface to make sense of real-time data from ~44,000 smart meter units across an energy grid — in collaboration with Thy-Mors Energi, RemoteGRID, and Kamstrup.
- Role
- UX Designer & Interaction Designer
- Year
- 2018
- Organisation
- Aalborg University
Outcome
Delivered a validated high-fidelity interface concept that reduced time-to-insight for grid operators, confirmed through usability testing with domain experts.
Three industry partners — Thy-Mors Energi, RemoteGRID, and Kamstrup — gave us access to a real problem: their grid infrastructure was generating enormous volumes of data from approximately 44,000 smart meter units, updated every quarter hour. The people who needed to act on that data — grid operators and analysts — had no interface designed for the task.
The challenge
Raw data at this scale isn’t a visualisation problem — it’s a sense-making problem. The goal wasn’t to display all the data. It was to surface what matters, when it matters, to the people who need to act on it.
Process
We used a mix of methods to understand the domain before touching any design tools:
Contextual inquiry — observing and interviewing grid operators in their actual working environment. What are they looking for? What decisions do they need to make? What do they currently do when something looks wrong?
Workshops — collaborative sessions with stakeholders from all three partner organisations to align on priorities and surface domain knowledge that wouldn’t emerge from interviews alone.
Participatory design — involving operators directly in sketching and critiquing early interface concepts, rather than presenting finished designs for approval.
Design decisions
The resulting interface concept prioritised:
- Anomaly detection over raw data display — the interface surfaced deviations from expected patterns rather than requiring operators to find them manually
- Spatial mapping — geographic context for where anomalies were occurring, not just that they existed
- Drill-down structure — overview → region → unit, with each level showing only the detail needed at that scale
What I took from it
Working with genuine industrial data at this scale made abstract interaction design principles concrete. “Progressive disclosure” isn’t a pattern — it’s the difference between a dashboard an operator trusts and one they ignore.