Di-BOSS: A Summary of the Research, Development & Deployment of the world’s first Digital Building Operating System
The Digital Building Operating System (Di-BOSS) was built as a three way collaboration among Rudin Management Company, Columbia Engineering’s Center for Computational Learning Systems (CCLS), and Selex ES, a global technology company owned by Finmeccanica, with each party bringing unique skills to the “3 Legged Stool.”
First and foremost, Rudin Management contributed their industry leading operational expertise in commercial real estate operations, and particularly their skills in energy efficiency to the Di-BOSS team. The system was built from the engine room out rather than from the current industry standard of from the top down. The latter is the specialty of companies that traditionally have sold software rather than operated buildings. Subsequently, the machine learning algorithms have codified these industry-leading innovations into Di-BOSS.
For over 100 years, Rudin has been the leading private owner/ manager of office space in Manhattan. They own and operate more than 10 million square feet in their 16 building commercial portfolio, and an additional 5 million sq. ft. of residential buildings. Several of their buildings are critical to the economic vitality of New York City, including perhaps the single most important building in the city, the original AT&T Long-lines hub for all trans Atlantic telephone cables. That building is now central to almost all cell phone, HDTV, and internet-cloud traffic between NYC and the rest of the world. Another of their buildings houses the world headquarters of Thompson Reuters news service in Times Square, and their flagship on Park Avenue houses the worlds largest accounting firm and the National Football League. Visit www.rudin.com.
Since 2004, Rudin Management has improved the energy consumption of their entire portfolio by more than 15percent for electricity, and by more than 40 percent for steam and natural gas (in BTU equivalents). They have won numerous national awards for both energy savings and environmental sustainability while lowering their overall energy costs from ~$60 million to less than $40 million per year. Year-over-year energy cost savings at Rudin have averaged 7 percent annually since 2008, with the Di-BOSS pilot in 2 million square feet saving $1 million of that since system testing began in the summer of 2012.
The Di-BOSS collaboration began with the American Recovery and Reinvestment Act of 2009, which was intended to act as a stimulus to the recessionary national economy via the funding of improvements in infrastructure, including the Smart Electric Grid. Consolidated Edison of New York (Con Ed) had been collaborating with Columbia University’s Center for Computational Learning Systems on Smart Grid research since 2005. While assembling a winning proposal for Department of Energy ARRA funding, Artie Kressner, director of R&D, and I approached Rudin Management for their expertise in smart energy innovations on the consumer side of Con Ed transformers.
We invited John Gilbert, COO, and Gene Boniberger, VP for operations, of Rudin to the Manhattan Electric Control Center to ask what information might be useful to further Rudin energy efficiency improvements. We showed a 40 second warning in voltage and frequency sags before the great northeastern blackout of 2003 that I had integrated from measurements of transmission voltage and frequency being tracked at two different consoles of the Con Ed bulk power control center. Artie asked Rudin if that short warning time could have been useful to Rudin. Absolutely yes was the simultaneous response from both John and Gene. Elevators could be ordered to safe exit positions if computers could communicate directly between Con Ed and Rudin’s elevator management system.
However, Rudin realized that no system-of-systems existed in their commercial building management systems that could cross all subsystem software packages required to safely and securely send such a warning signal. Two patents were filed by Rudin and Columbia: one for the system-of-systems invention that first described what became the industry’s first Digital Building Operating System, Di-BOSS, and the other for the system for two-way instantaneous communications of electric grid power quality on both the utility and consumer sides of the transformer.
These patent were combined with several Columbia Con Ed adaptive stochastic control patents to become a cornerstone of the DOE Smart Grid proposal submitted by Con Ed in 2009. Though the DOE award was won in 2010, Con Ed used a “termination-for-convenience” clause of the joint contract to quickly turn to a different application of customer/utility systems collaboration that formed only a small component of the overall Rudin and Columbia vision, automated demand/response using a third party intermediary.
In late 2010, Rudin and Columbia decided to pursue what would become Di-BOSS on their own, and Columbia began prototyping of the Total Property Optimizer (TPO), the Machine Learning (ML) system, which would ultimately become the brain of Di-BOSS. Columbia put “boots-on-the-ground” at Rudin, and I basically moved into their 345 Park Ave. flagship building (a 1.8 million sq. ft. office tower). Rudin management and engineers described in detail what they needed but could not find in the marketplace, and Columbia built a prototype to those specifications.
During this period of prototyping, the Rudin team would pour over each new ML and visualization capability, critiquing and correcting until the TPO was accurately forecasting electricity and steam consumption for each coming day of operations. From pre-heating in the winter, to startup each morning, to heating-up and cooling-down and ramp-down at the end of the day, recommendations were manually made by the TPO and acted on by the Rudin engineers, and energy consumption improvements tracked. Meantime, Rudin was outfitting key locations within 345 Park Ave. with temperature sensors on tenant floors, and expanding the pilot project coverage to 560 Lexington, a 300,000 sq. ft. building around the corner.
In March 2011, TPO began forecasting space temperatures for tenants at 345 Park, and by November, at 560 Lexington. The machine learning uses past history for each day of like weather conditions, along with the future weather forecast for tomorrow to compute most likely conditions and timing 24 hours ahead, and runs a new Support Vector Machine prediction every hour. Later on, when Rudin wanted shorter term forecasting so that they could steer the building during rapidly changing conditions, Columbia invented a 2-hour-ahead forecasting system called “Now-Casting.” Better overall building performance resulted from each new step in this prototype discovery phase of development.
By June 2012, Finmeccanica, Italy’s second largest conglomerate, was recruited to join the team because Rudin and Columbia knew they would need a world-class systems integrator to market Di-BOSS worldwide. Di-BOSS required that all the data coming from disparate legacy sub-systems feed data to a middleware database, and distribute it to TPO. From elevator management to the fire alarm system, security and building management systems, software systems would differ by manufacturer and vintage from building to building. A true middleware pro would be required, as well as a company with global marketing expertise if Di-BOSS were to become as ubiquitous as operating systems in computers, smart phones and iPads. Collaboration and operating contracts amongst the team were signed in October 2012.
The Finmeccanica electronics, systems integration and security subsidiary Selex EG was chosen to be the point for the collaboration. Selex EG had already marketed a systems integration facility (SIF) that would become the middleware that gathers all the data from the disparate sources, historically archives it, and feeds the TPO and other alarming and recommending components of Di-BOSS in real time. Rudin and Columbia combined their intellectual property and jointly licensed a total of 17 background patents and all future intelligent building patents developed by the team to Selex ES, splitting future royalties 50/50.
Selex ES immediately assigned a team from their Genoa headquarters to integrate the TPO prototype into an industry-hardened product. Amazingly, only eight months later, Di-BOSS was announced as a commercial product at RealComm’s Intelligent Buildings convention (IBcon) on June 10-12 in Orlando, FL.
For Di-BOSS to reach market so quickly was a remarkable accomplishment of teamwork among the three disparate teammates. The American subsidiary Selex Elsag NA was chosen to be in charge of sales and development of Di-BOSS in the North American market, and Selex ES would handle the sales and marketing for the rest of the world from Genoa.
During the pilot project at the two Rudin buildings, savings were realized across the board from two buildings that had already increased their energy efficiency by 30 percent since 2005. Return on Investment was realized from better building and tenant practices, additional efficiencies in OPEX and CAPEX, and replacement of reactive with anticipatory maintenance.
The Di-BOSS real-time re-commissioning involved the elimination of pre-start energy penalties from the utility, shortening the start-up times to fit the thermodynamics of each building that the Di-BOSS machine learned. Real time re-commissioning also provided better control of heat-up or cool-down, continuous steering to lease targets, investigation of maintenance issues before tenants become aware of the problems, better ramp-down controls tied to direct measurements of building occupancy, and detection and elimination of energy waste during overnight hours when the buildings were shut down.
Specifically, Rudin saved $505,000 during the winter of 2012-2013 by implementing the following strategies identified through Di-BOSS at 345 Park Avenue:
• $175,000 from preheating before startup in the morning by pumping 70,000 gallons of hot water heated before the 6 a.m. to 11 a.m. high steam demand time window set by Con Ed. This is a form of hydro-battery,
• $100,000 from an average of 30 minutes later startup recommendations each morning by Di-BOSS,
• $50,000 from better steering out of warm-up each morning,
• $72,000 from continuous re-commissioning of tenant space temperatures using the TPO Now-Cast to steer within the comfort horizon indicator of Di-BOSS,
• $50,000 from an average of 20 minutes earlier ramp-down times as tenants leaving the building before the shutdown time obligated by the leases,
• $23,000 from prevention of winter overheating of tenant spaces, and
• $35,000 from improved tenant base load electricity reduction overnight.
Electric consumption disparities among multiple tenants were identified. In particular, peak daytime highest usage to nighttime lowest usage ratios were targeted for improvement. Tenants were uniformly supportive of the effort, and were immediately brought in to help design the Di-BOSS tenant portal.
In summary, Selex ES is marketing Di-BOSS globally. Rudin has commissioned the installation in the two pilot buildings, and is rolling Di-BOSS out to six more office buildings in 2013, and eight more in 2014, including two large residential properties. By 2015, the entire Rudin commercial portfolio and the largest of the residential properties will be running on the Di-BOSS brain. Simultaneously, Selex ES, Rudin and Columbia are jointly designing and integrating the next generation of Di-BOSS with TPO into what will soon become Di-BOSS version 2, which will assure leadership in the exploding marketplace for Intelligent Buildings far into the future.
About SELEX ES
SELEX ES, a Finmeccanica company, is an international leader in electronic and information technologies for defense systems, aerospace, data, infrastructures, land security and protection, and sustainable solutions. From the design, development and production of state-of-the-art equipment, software and systems to through life support, Selex ES partners with its customers to deliver the information superiority required to act decisively, complete missions, and maintain security and protection for operational effectiveness. Selex ES is an integrated global business with a workforce of approximately 17,700 and total revenues in excess of €3.5 billion. With core domestic operations in Italy and the UK, the company also has a strong presence in the United States, Germany, Turkey, Romania, Brazil, Saudi Arabia and India. For more information, see www.selex-es.com.
About Rudin Management
The Rudin family has owned New York City real estate for more than 100 years. Family-run since its founding, the family’s real estate holdings rank as one of the largest and most respected privately owned portfolios in New York City. Among its holdings are 16 office buildings containing approximately 10 million square feet of space and 21 apartment buildings comprising more than four million square feet of residences. The Rudin family is committed to developing sustainable real estate that is respectful of its environment and surrounding community. For more information, visit www.rudin.com.
About Columbia Engineering
Dr. Roger Anderson’s team at the Center for Computational Learning Systems in the Fu Foundation School of Engineering and Applied Science of Columbia University encompasses exploration of next generation software and machine learning systems to control electric grids, manufacturing operations and the recharging of fleets of electric vehicles. His team specializes in the smart grid, smart cities, integration of control center operations for energy companies, real options and portfolio management, 4D reservoir management and shale gas and oil production within the oil and gas industry, as well as alternative energy research including solar, wind and large industrial battery production and storage. See http://eesc.columbia.edu/faculty/dr-roger-n-anderson and http://ccls.columbia.edu. Anderson’s team is also affiliated with Columbia’s Institute for Data Sciences and Engineering and Lamont-Doherty Earth Observatory. For more information, see http://www.engineering.columbia.edu.