Behind the glittering, sculpted glass skin of the San Francisco Public Utilities Commission's new 13-story headquarters beats the heart of one of the most energy-efficient office buildings in the world.
IT makes it tick.
The office tower, a smart building that opened in June, features many state-of-the-art green technologies, including solar panels and wind turbines that supply up to 7% of its energy needs. "Our goal was to save rate payers money and to educate them about energy efficiency," says program manager Masoud Vafaei. A LEED Platinum candidate, the building uses 32% less energy than similar structures with conventional designs. Like many smart buildings, the SFPUC headquarters uses computer technology to manage and optimize the many systems that control every aspect of the building's operation. But much of the day-to-day efficiency gains are derived from a central database that pulls data from all of the building's management and control systems, and from the use of analytics to study that data to ensure that all systems work in concert to minimize energy consumption and optimize operations.
Such IT-driven designs can also be applied to existing buildings, generating energy savings of 5% to 10% simply by optimizing how existing systems run, experts say. "There's a huge opportunity for building owners to do the sorts of data mining that other industries have done for years," says Jim Sinopoli, managing principal at Smart Buildings LLC, a design, engineering and consulting firm. "Using analytics, you can predict when there's going to be a failure and when to do preventative maintenance."
Over the past several years, building management and control systems have been gradually converging with traditional IT infrastructures. Open standards now dominate at the hardware layer, where industry-standard communication protocols allow data collected by data points such as sensors and valves to flow over the corporate IP backbone to server-based building management systems that control everything from heating, ventilation and air conditioning (HVAC) to lighting, power, fire protection, security, elevators and building access.
"Sensors are becoming better, smaller and, most important, cheaper," says James Dagley, vice president of marketing at Johnson Controls, which sells an automation system used in the SFPUC building. For example, the vendor is testing a technology that uses Ethernet to power LED lights, each of which includes 12 different environmental sensors that measure things such as light, humidity, temperature and motion. "Each sensor costs about two cents," says Dagley. All of that data can be tied back to a Johnson Controls building automation system in the San Francisco office tower.
"From a hardware standpoint, the industry has fulfilled its goal of integrating," says Tom Hartman, principal at The Hartman Co., an engineering firm specializing in smart buildings. "But from a software standpoint, there hasn't been much progress."
By the Numbers
The SFPUC's Green Headquarters
Square footage: 277,500
Height: 13 stories
Construction cost per square foot: $257
Annual energy use: 2.8 million kwh
Renewable energy sources: Solar panels, wind turbines
Maximum energy produced by renewable sources: 227,000 kwh per year
Percent of total energy consumption supplied by renewable power: 7%
Projected overall energy savings versus conventional building: 32%
Expected life of building: 100 years
Projected energy savings over the building's lifetime: $3.7 billion ($500 million in 2012 dollars)
ROI: 26 years
"When it came to pulling all of that data into one platform to streamline management of the building, that wasn't available," Vafaei says. So the commission developed an integrated building management system (IBMS), a custom-built SQL Server database that pulls data from every monitoring and control system, including those that regulate heating and lighting, elevators, generators, solar arrays, the internal window blinds and external shutters that adjust natural lighting, and the roof-mounted weather station. "The IBMS provides a management layer on top of the traditional controls," Vafaei says. The system also aggregates data and provides information dashboards that give an end-to-end view of all systems to building managers, executives, employees and even the public, by way of a 40-foot-wide media wall in the main lobby.
Smart Buildings' Sinopoli worked on the IBMS. "We're at the point now where you can integrate these building systems. An IT infrastructure has really penetrated all building systems," he says. And once the data has been integrated, all of those systems can be functionally connected so that an event in one can trigger a response in another.
At the SFPUC building, for example, the IBMS applies real-time analytics to data from the shade, lighting, HVAC, weather station and room occupancy sensors to determine how shade positioning will affect both cooling and lighting system loads. The shade position is then adjusted automatically.
Not Just for New Construction
Existing buildings can also benefit from an IBMS, says Darrell Smith, operational supervisor at Microsoft's Real Estate and Facilities organization. The company's Energy Smart Buildings project, now under way in the 118 buildings that make up its Redmond, Wash., campus, uses an IBMS and analytics tools to optimize operational and energy efficiency across seven building management systems. The IBMS pulls data from those systems, which track HVAC, lighting, power monitoring meters, generators, power distribution units and circuit monitors. Because of the complexity, says Smith, partnering with IT was critical. "They looked at the protocols with us and how we were going to get the data out of these systems," he says.
Microsoft's campus has 2 million mechanical and electrical data points (the SFPUC building has 13,500) that generate 500 million data transactions, or data point updates, per day. "The business was doing nothing with that data," Smith says. Replacing those building systems, from power metering to lighting and HVAC, would have increased efficiency at a cost of $50 million to $60 million. Instead, the facilities group decided to extract the data from its existing systems and transfer it into a common SQL Server database, where the data could be analyzed and each building's operational performance could be assessed using key performance indicators, such as power demand per person and average demand per square foot, as well as a building performance indicator rating for each type of building (a lab or an office, for example). Microsoft generated operational dashboards for facilities staff, and it will soon offer plug-level usage data for its Sustainability Champion program, which will let employees see how their individual energy conservation efforts pay off.
Sometimes Fingers Are Better Than Sensors
There are a number of high-tech tools that can help organizations save money and cut energy use by automating building controls, but sometimes the best option is a manual switch.
Consider the case of Union Hospital in Terre Haute, Ind. When officials wanted to save energy in operating rooms, the first thing they did was to stop relying on occupancy sensors to control the air conditioning.
That move might seem to be counterintuitive. Operating rooms require 15 or more air exchanges per hour when in use, and turning down the air conditioning when a room is unoccupied can save a lot of energy -- potentially cutting costs by as much as $10,000 annually per room. The problem was that hospital staffers move in and out of the surgical suites all the time, so the occupancy sensors, which worked fine for lighting, couldn't efficiently control the ventilation system.
CIO Kym Pfrank's next move was to try a product called the Healthcare Environment Optimization (HEO) system from building automation system vendor Johnson Controls. HEO can be integrated with a hospital's surgery scheduling system, and Union Hospital could have configured the software to turn the air conditioning in surgical suites on and off in concert with the surgical schedule.
But that wouldn't have worked either.
"The schedule flexes too much, so you still need a [manual] control" in each room, says Pfrank.
Another option would have been to take advantage of functionality in HEO that activates a room's air-conditioning system when a patient's ID wristband is scanned. But in the end, the surgical teams said that they preferred to activate and deactivate air conditioning manually using the system's touch-screen monitor. Now, a staff person touches the screen when a procedure is about to start and then touches it again when it's over, so HEO can throttle back the air exchanges to five per hour.
"We're still dependent on the human element," says Pfrank. "I don't know that that can be ever taken fully out of it."
-- Robert L. Mitchell
Both energy-efficiency optimization and fault detection and diagnostics are based on rule sets that are typically customized for each project. The rules determine whether equipment is operating efficiently; if there's a problem, the system performs tests to find the cause.
The SFPUC's rules, which were developed using a spreadsheet-based energy analysis tool called eQuest, also calculate the increased cost associated with running a system out of specification. When a critical event occurs, the IBMS can automatically generate a work order in the facilities management system, says Sinopoli.
Microsoft's engineers created 195 rules and used SQL Server's Stream Insight event engine, along with analytics software from Iconics, to perform calculations that identify faults and monitor efficiency. "We not only find the faults, but monetize them," says Smith. For example, variable air volume (VAV) boxes control airflow in the air conditioning system. If one of the 20,000 VAV units isn't properly calibrated, the system alerts the facilities group before any employees call to say they're uncomfortable. The rules also calculate the energy cost savings that would result from fixing the problem, allowing the facilities group to prioritize the work. "We went from walking around to figure out what's not working to figuring what's not working and costing the most. That saved us over $1 million right there," Smith says.
The system also has allowed Microsoft to start moving toward a just-in-time maintenance and tuning schedule, a trend known as continuous commissioning. Following a traditional maintenance schedule, more than 26,000 filters would be changed quarterly, and each of the more than 800 building air-handling systems would be tuned in a five-year rotation. With the new system, Smith says, "we were able to go much deeper with the data and tune all [30,000] of the assets, not just the large building systems." The problem with tuning 20% of the systems each year is that, as with cars, the efficiency and performance of building systems degrade gradually over time. Now Microsoft uses analytics to replace each filter based on actual usage.
"Instead of changing them on a schedule, we change them at the right time. That's the intelligence we're talking about -- a building generating its own work orders," Smith says. And by prioritizing maintenance needs, the facilities organization can continually tune the campus. "It compresses the five-year cycle into one year for a total savings of $1 million," he adds.
The Redmond campus project, which is about 20% complete, has also allowed Microsoft to reduce its peak energy demand. "We were causing our own peak demand just by how things were occurring in the building," Smith says. Resequencing when different building systems came online smoothed out the demand curve. In the pilot phase, Microsoft has so far shaved energy costs by 6% to 10%, while the application of analytics for fault detection and diagnostics is projected to save more than $1 million annually. "Our payback on this will be about 18 months," he says. That payback period is shorter than it would be in other states, however, because Washington has the country's third lowest electric power rates.
Condition the People, not the Building
The best-laid plans for constructing smart buildings often fall victim to poor processes, contends Tom Hartman, principal at The Hartman Co., an engineering design firm. Take, for example, how buildings are heated and cooled. Most systems focus on conditioning the building envelope, or areas that include a half-dozen or more office areas. What's more, many systems are still designed to wash air around the exterior walls because that's where most heat loss used to occur.
"We need to change the philosophy from conditioning the building to conditioning the people wherever they are," says Hartman. Low-cost sensors and wireless networks can help make that possible. But the number of sensors -- measuring occupancy, CO2, temperature, humidity, light and more -- would need to be vastly increased, to an average of up to 12 per occupant, he says.
While the ability to create integrated, all-in-one sensors will help keep costs down, the increase in the number of data points that need monitoring would increase from an average of 3,000 to 4,000 in a large building today to more than 20,000 for a building with 1,000 occupants.
"It's not terribly difficult or expensive to do," says Hartman. "But better database management has to occur. That's the key."
-- Robert L. Mitchell
On the Leading Edge
Bringing this level of integration to smart buildings is still considered leading edge. "Analytics for fault detection are pretty new on the market, and Microsoft is the first to apply diagnostic and fault detection technology on a mass scale," says Russell Agrusa, president and CEO at Iconics, which develops and markets software for factory and building automation management. A building needs to be at least 100,000 square feet to make the investment pay off, and it should have metering at least on every floor to take full advantage of the technology, he says.
Several vendors, including IBM, Microsoft and Tridium, offer tools to help with integrated, data-driven designs that provide end-to-end views of building control systems. But achieving this level of integration in new construction isn't always easy, says Rawlson King, spokesman at the Continental Automated Buildings Association in Ottawa, Ontario. "You can still contract with integrators and not have the systems working to your specifications, despite the fact that they are experts," he says.
One reason is that traditional construction processes aren't optimized for this level of integration, and the participants -- the architects, engineers, general contractors and subcontractors -- tend to operate in silos. The architect, design engineer and general contractor aren't in communication with the people doing the actual work -- or with one another. Instead, each reports to the building owner, and each has different priorities, Hartman says.
"Almost every chilled water plant in the U.S. today has been designed, programmed, commissioned and operated by entirely different sets of individuals who rarely talk to each other," he says. As a result, most operate at double the energy consumption that's necessary. "Because of that, it's been impossible to raise the level of sophistication of controls. It's not better technology or products. It's the process," Hartman adds.
The SFPUC broke with that tradition and took more of a team approach. "We were fortunate to have a design team and a construction team that worked collaboratively. That was paramount in making this a success." Vafaei says. But, he acknowledges, "it took time for everyone to get to that stage."
Vafaei also recommends having vendors sign a compliance statement indicating that they support open communication and database protocols. This ensured that the commission could pull the data it needed from every system.
"When it comes to controls and automation, everything should be open. As we replace things, that's where we're going," Smith says.
Both Microsoft and the SFPUC also enforced the use of a naming convention for data points on all control systems so as to avoid a name conversion step when importing data into the IBMS. Microsoft had the foresight to require vendors to use its 32-decimal naming convention years ago; the public utilities commission required every vendor to use a convention it published.
Things aren't perfect, though. Microsoft still needs to automate reporting. For example, the process of reading more than 1,000 meters, normalizing that data and getting it into the database is still manual, and the tools for managing a smart building holistically are still evolving. But the industry may finally be at a turning point.
Ten years ago, building automation consisted of using dial-up connections into PCs, one for every system, Smith says. "But in the last couple of years, we've jumped the chasm," he says. "Leveraging IT to optimize smart buildings is here."
However, he adds, that level of "personal control" is still at least three to five years away from mainstream adoption.