Designing for the people who run wind farms

A software company whose platform lets control centres monitor wind and solar farms and operate their equipment in real time: turbines, inverters and substations. Automating those operations was the heart of the product.

I joined in March 2020, at the start of the pandemic, and worked there until May 2021. The move to remote work was not planned, so my onboarding was access to a test environment, and the developers were the ones who helped me understand what I was looking at.

The work ran on three fronts at once: improving the product that already existed, designing tools for individual clients and designing its successor.

The work, in order:

1. 01Audit33 screens of the existing product captured, and its navigation rebuilt as a map.
2. 02DomainWhat the data meant and how operators worked, learned from developers, an engineer and clients.
3. 03Client toolsShift management for a control room, on-call planning and a notebook for field technicians.
4. 04Design systemAtoms, molecules and organisms, built side by side with two frontend developers.
5. 05Substation diagrams20 diagrams of 13 substations redrawn with one library of symbols.
6. 06RoadmapA year of design in 26 two-week sprints, with the same four steps for each of its 17 features.

And the decisions that shaped it:

• In the diagrams, the symbols stay and the presentation changes

  A substation is supervised on a single-line diagram: an electrical drawing with a symbol for every transformer, line and switch. Making it look modern by redrawing those symbols would have changed what operators read at a glance. I worked with an engineer to keep the meaning of each one and adapted only how they sit on a screen: one library of symbols with their states, open or closed, selected or not, and a text style of its own. Operators in the company reviewed the result.

• A design system built around the table

  The product was inconsistent from screen to screen, and its frameworks decided how things looked. A system of our own fixed the first and freed us from the second. I organised it in atoms, molecules and organisms, and its biggest piece is the table cell, with 119 variants, because the product is made of tables of live data. Every flow went to development with each screen marked with the component it used and what every control did.

• A field notebook that works without coverage

  Field technicians wrote their inspections in a paper notebook and typed them into another device later. I designed a digital notebook with the checks to run and fields for the values measured on site. Wind farms have poor coverage, so the design keeps the data on the device during the visit and synchronises by itself when the connection returns. Asking technicians to wait for signal was never an option. It was built for the client that asked for it.

• One process for every feature, written into the plan

  In January 2021 I planned the year of design: 17 features of the new product spread over 26 sprints. Each one follows the same four steps, analysis, desktop prototype, design system documentation and interaction documentation, with three fixed checkpoints: a review meeting, changes and sign-off, and a deadline. Documentation is a step of every feature, not a task for the end.

In a technical domain, modernising an interface starts with understanding the conventions of the people who use it, and keeping them. Working with engineers and operators taught me to tell a change of presentation from a change of meaning. I also learned that a client’s need can open a design conversation that an internal proposal cannot.

Today I would agree a shared way of working, the criteria to prioritise and a base of components much earlier. What I cannot claim is fewer errors or faster work in the control room: I saw parts of these interfaces in production, but nobody measured them, and I left before the new product was finished.