Blind data: solving smart buildings’ UX problem

The world of buildings is becoming as much about data as it is about physical materials. As humanity battles to halt climate change and find new ways to exist without further damaging the environment, we must find better ways to manage energy consumption. Technology can help us make buildings and cities more intelligent and more efficient — but only if we can actually use the digital information that it generates. 

This is at the beating heart of the net-zero challenge. Even with all of the data in the world, without effective ways to manage and understand it we will be unable to apply it to deliver any form of real-world impact. 

Data collection can be considered somewhat of a solved problem in buildings, at least as far as technology is concerned. Most buildings are not yet equipped with a vast array of sensors, but there is already a wide range of wireless and wired devices at every price point. Whatever physical characteristic is of interest, there is likely a sensor available to measure it, at a steadily declining cost — driving an explosion in internet of things (IoT) applications. 

But this only represents the first of three stages that building designers and operators need to pass through in order to get to the point of actually delivering operational benefits. The next is to manage that data in a way that makes it easy to access and to analyse, and the final — crucial — step is to understand what the data is telling us and use this insight to inform better decisions, either through manual intervention or automation.

Why does it matter how we manage building data? Compare it to poor labelling and formatting in a spreadsheet. Without consistent column and row headers or contextual information, the value and usability of the data contained within that spreadsheet becomes extremely limited. Similarly, poorly managed building information is liable to be under-used or, worse, not used at all. It will be expensive and time-consuming to analyse, and very difficult to scale, to reuse or to apply in new situations in the future — posing a challenge to future smart cities.

These second and third stages remain largely in development, especially regarding the cost-effectiveness of currently available solutions. The importance of good information management has been acknowledged by several industry and academic projects, such as Project Haystack, Brick Schema and RealEstateCore, all three of which have seen reasonable adoption in the market. These provide a technical structure to organize building information — a shared language that can be used by different systems to describe objects within a building and their relationship to other things.

Valuable though these projects are, there remains a significant separation between our ability to pull information from a building and our ability to use it. Central to this is the skills gap that exists between the data scientists who are comfortable interpreting complex information and the people who are actually operating the building. Information management is a very technical exercise and, to date, it has been approached from the perspective of a building designer or engineer. It’s very logical if you have the building drawings and you know that you have to connect a duct to a VAV box to the hot water supply and the cold water supply. But if you’re a property developer or a facilities manager who wants to know how effectively you’re heating the second floor, and who doesn’t have this contextual information or experience of interpreting complex datasets, it’s not at all intuitive to work out what’s going on.

Fully autonomous buildings are still on the distant horizon and human intervention will be the reality for the short-to-medium term — so it is imperative that we find a way to merge the digital components of smart buildings with the manual components of building operation. I began looking at this for my Masters thesis at the University of Cambridge, and it’s the subject of the PhD that I’m studying for now. The obvious, perhaps cliched, answer is to adopt a more user-centred design approach. We need to rethink the interfaces between human and machine, focusing on how we deliver information to building operators, what this information tells them, how they can use it and how they can feedback their own intuition and knowledge to the system. We need to start from the perspective of someone using a building and think about what they want to know, and then find a way of describing and connecting that information.

This is a new way of thinking about design, but we don’t have to start with a new building. It would be another, cloud-based layer sitting on top of an existing building management system, connecting the dots to support existing buildings to be more efficient. The things a designer might want to know are far more predictable than the questions that a future user might ask. So we have to build in flexibility, so a data model is designed in such a way that you can push one domino and the others fall into place to help you find the answer. At a very simplified level, it’s almost like the names we give the rows and columns in a table. At the moment, these only make sense to certain people and they’re in an order that only makes sense if you already know how to analyse it. There is no column for “temperature” or “energy”. So we need to find a way of mapping that backstage data to sit within a far more transparent interface. What if instead of having a table, you break it down into a spider’s web of cells that are connected to each other in a more fluid way? This is the end goal of my PhD: to build a prototype of that web, defining the rules and relationships, and validating it to ensure it’s as scalable and as useful as it could be.

Ultimately, a lot of sustainability is about how we use buildings and how we employ data within them. But data scientists don’t usually run buildings. This is about giving the tools to the people who do.

James Kinch is a smart technology consultant with WSP, based in the UK