The machines are coming – indeed, they are already here – and they are talking to each other. But rather than being the villains in a science fiction movie, these machines are here to help. They connect our homes, cars and workplaces to the Internet in the quest to make life better.
In late 2013, the research firm Gartner estimated that the “Internet of Things” – the vast network of automated pieces of hardware that are individually connected to the Internet – would consist of 26 billion devices by the year 2020, a number that does not even include laptops, tablets and smartphones (another 7.3 billion). Hundreds of millions of these devices will be located in buildings.
Facilities are only beginning to harness the possibilities of such incredible interconnectedness. Arrays of sensors on building systems like HVAC and lighting are now part of the Internet of Things. They are not just monitoring status and alerting facilities to malfunctions; they are constantly capturing data and sending it to the cloud. But managing the sheer volume of this data presents its own challenge.
“As you add more sensors, you add more power,” says David Kollmorgen, an international director with JLL who leads the firm’s business intelligence practice. “But you also add a lot more data. Each layer makes the data set grow exponentially, not linearly.”
The usefulness of this data extends beyond variance analyses and if/then scenarios, Kollmorgen adds: “What we’re starting to see is data scientists looking at these huge clusters to identify patterns that we haven’t even thought about yet.”
This is the true value of “big data” – a collection of information so vast as to be indecipherable without the aid of advanced computer-driven analytics. Facilities are just approaching the frontier of these capabilities. But a cottage industry has already developed to help companies capitalize now. Learn how you can take advantage.
Measure More to Manage Better
“We are taking big data in the direction of managing operations and utilities,” says Logan Soya, founder and CEO of Aquicore. Soya’s company uses web-enabled sensors to rapidly collect consumption and equipment operation data.
“Buildings are going from a dozen data points on a utility bill to real-time data reported every 30 seconds,” he says. For a simple building, that extrapolates to over 12 million data points in a year. For a more complex building with multiple submeters or a large portfolio, the figure quickly escalates into the billions.
“The sensors are BAS-agnostic,” explains Soya, “and it’s a cloud-based platform, which means a manager at any building can have decision-making data within a few hours of installation.”
For one client, that data revealed that the energy consumption curve of one building showed an unexpectedly high base load. Further analysis then flagged the heat pumps as the source of the abnormality. Upon examination, it was discovered that the relays between the pumps to the BAS were not connected properly, which meant they had been running continuously at night despite the indication from the BAS that they were shutting down as scheduled.
“The solution was simple and cost practically nothing,” notes Soya, “but it would have taken forever to find without the analytics.” As a result, the building will save $80,000 in annual expense, far exceeding the upfront cost of the sensors.
This data has tremendous value even when there is no obvious malfunction to correct, Soya adds.
“We have seen buildings raise their ENERGY STAR ratings by 50 points and save hundreds of thousands of dollars as our tools help them optimize scheduling,” Soya says, noting that the value extends even further when the data feed indicates increased stress on equipment, often indicating an impending shutdown.
“It’s like giving a car a full dashboard when it has never even had a speedometer before,” Soya says. “A lot of buildings are still working with static monthly bills, so there is a lot of opportunity out there.”
A Few Degrees of Freedom
The usefulness of BAS data doesn’t stop at tracking energy consumption to learn from the data feed. In fact, some systems flip that paradigm and learn from their human operators.
“Machine learning is a big part of the story. There is a lot of repetition and pattern recognition involved in looking at data, and it just isn’t scalable,” says Lindsay Baker, VP of business development for Building Robotics, a building controls company that grew out of UC Berkeley’s computer science department. The firm’s breakthrough product is Comfy, an interactive temperature control system built on a machine-learning platform.
Comfy works in a counterintuitive way – it gives individual workers a measure of control over the temperature in their space. Workers are given access to a web portal or iPhone app that allows them to dispatch one of three messages: “Warm my space,” “I’m comfortable,” or “Cool my space.” This control, which is connected to the BAS on the back end, responds immediately to the request.
“If someone wants cooler space, that person’s zone will get about 10 minutes of cool air,” Baker explains. “People are usually amazed at the response, which makes them feel like the building is listening to them.”
Critically, the system does not immediately alter the set points. Instead, it assimilates the feedback and begins to learn. Over time, it will widen dead bands in zones with little feedback, especially unoccupied zones like elevator lobbies and copy rooms. In areas where feedback is regular, it will optimize to a narrow range of temperatures based on worker preferences – and it will adjust those ranges as preferences change over the course of days, weeks and seasons. It also learns to widen bands on days when zones are unoccupied, like weekends and holidays.
At one of Building Robotics’ first installation sites, the client measured a 23% reduction in energy use for heating and cooling over the six-month pilot period. And, in a development that was perhaps even more appealing to the facility management team, hot and cold calls were reduced to zero. “The complaints are a big headache for facility managers, so they didn’t mind giving occupants some freedom if it made them happier,” notes Baker.
For her part, Baker has gained new insight as well. For instance, people in warmer climates tend to gravitate toward an optimal temperature that is closer to the high end of the range set by facility management. Users also generally want their space warmer in the morning and cooler in the afternoon.
“People are unpredictable,” she says. “The notion behind Comfy is to embrace that unpredictability and give them some control, while at the same time not over-conditioning the space.”
Know Your Space
The same concept behind optimizing the temperature in a workspace applies to optimizing the use of the space itself.
“We’ve been talking for 20 years about optimizing utilization,” says Tony Booty, a director at Abintra, a UK-based firm specializing in workplace measurement. “The difference is that now we can actually do it.” His statement is based on the convergence of a truly mobile, flexible workforce enabled by wireless connectivity and WiseNet, his company’s real-time utilization measurement product.
WiseNet’s sensors continuously capture seat-level occupancy with passive infrared devices mounted inconspicuously under desktops and conference tables. Whenever they detect motion in the human temperature range, they transmit messages over a network via simple wireless radios, eventually creating a data history that reveals which work areas are used at what times. This helps FMs understand how to size and configure spaces for optimal utilization. Occupancy rates can also be displayed on a three-dimensional location grid for quick reference.
To date, Abintra has measured over 100,000 workstations and collected over 3 billion data points. In doing so, the company has helped clients optimize to their own reality, not just a design specification. “Our tools don’t just tell companies they are wasting space,” Booty explains. “They identify which space is not being used, when it is vacant and what type of space it is.”
Booty is quick to point out the payback on the sensors, each of which costs less than 1% of what companies typically spend annually to provide a workstation. One of Abintra’s success stories involved measuring a client’s space in advance of a planned relocation. Leveraging its actual utilization data as measured by WiseNet, the company’s new space was configured with 46% fewer workstations and three times as many collaborative areas. The result was a 7% increase in workstation utilization. Another client saved more than $6.7 million by eliminating 300 workstations and altering its desk sharing model. But reducing space is not the only valuable outcome of measurement.
“We had a client whose employees expressed some privacy concerns (see sidebar on page 36), so we ran a pilot program on one particular wing where the workers all agreed to be measured,” Booty recalls. “As the data came in, the client updated the layout and furniture in the wing.”
The employees reacted positively to the new space (much to the envy of those not participating in the pilot, who soon agreed to be measured themselves). “In a case like this, the actual amount of space may not change, but people appreciate having more flexibility in the type of space they have,” Booty adds.
Coming Soon to a Building near You
The benefits of big data for commercial facilities are already tangible, even if relatively few companies are currently capturing them. But the next wave is already forming, and those on its edge know where it is going to crest.
“We’ve moved from reactively fighting fires to delivering a controlled response,” says Soya. “The next level is going to be predictive modeling.”
Using analytics to build accurate forecasts of building conditions is the next key to unlocking big data’s value. Such models have powerful implications for running buildings efficiently.
“With a couple of years of utilization data, we can build a model for load in a building,” notes Kollmorgen. “We can then tune the building’s systems more closely to its true operating needs.”
Predictive models also hold promise for better equipment maintenance and capital budgets. What if, for example, engineers no longer needed to rely on voluminous preventive maintenance libraries in order to get the most out of their equipment?
“If we know, for example that 5% of one type of equipment fails after 36 months, but 70% fails after 48, we can repair or replace them at 42 months,” Kollmorgen suggests. “Big data offers us the chance to find the patterns and identify the correlations and causes of malfunction.”
Such are the possibilities of a connected world, where smart phones, smart buildings, and even smart cities are producing unprecedented volumes of information. Already the value is real. As more and more data science is applied to buildings, FMs can look forward to even more innovative, dollar-saving solutions.
Phil Mobley is principal of Koine Communications, a content strategy and development firm that serves the commercial real estate industry. Phil can be reached at firstname.lastname@example.org.