Genentech/Roche has innovation in its DNA, and is now showing all the markers of a leader in data-driven strategies for facilities management and building operations.
Genentech/Roche cut the ribbon on its latest building in South San Francisco on May 21st, with California Governor Jerry Brown holding the scissors. The Governor came to call attention to how such private companies are designing the way toward a more sustainable energy future. Weeks earlier, Paul Van Buskirk, Senior Manager of Facilities Data Technologies, took to the stage at the OSIsoft User Conference to explain Genentech’s approach to managing all its properties, which now comprise about 8 million square feet of manufacturing operations, research laboratories, and office space just in the Bay Area, including this new 255,000-square-foot 7-story LEED certified office tower.
The stringent process requirements of the tightly regulated biochemical/pharmaceutical industry as well as Genetech/Roche’s commitment to sustainability have pushed the biotech pioneer to the forefront of operational IT. The Genentech Facilities Data team manages over $25 million in automation assets across two campuses and is dedicated to reaping ever higher levels of value from all the data streaming from its building automation, information technology and control systems. It uses OSIsoft’s PI system to centralize and analyze all this data. Van Buskirk explains:
“At Genentech, we subscribe to the principle of using one set of numbers to run the business. In 2013, we launched a project to better support that principle when it comes to facilities monitoring. We wanted universal data trending and a historian across all systems that impact our product, people and processes. Also, we wanted more robust and reliable alarming to reduce our risk exposure.”
Genentech’s buildings house a wide diversity of functions — research, production, clinical work and business executives and staff. Each of these functions pose different requirements to process control equipment, building management systems, lighting control systems, and other infrastructure.
PI is integrated with a visualization system for alarming, from OSIsoft partner Zymergi. It trends data from a branded building control system that is one of the largest deployments of this particular BMS with 350,000 points managed. Three other BMS systems also send data to PI, as well as two different lighting control systems and an automated solar shading system. PLCs are pulled into the PI system as well. along with wireless energy meters that have been strategically placed throughout the mixed-use buildings to monitor temperature as well as energy consumption.
“The compression engine in PI is good. Also, it is a low-bandwidth utilization system. You have to consider these storage and communication system costs too,” explains Van Buskirk. Due to limited internal storage, a BMS system will conventionally save trend data for a limited period, for example, two-weeks or a month. The PI historian saves trend data for as long as it might be useful, potentially years. Van Buskirk further explains that the BACnet protocol can be broadcast-heavy, using polling to check in with connected equipment. “With PI, we can set communications up on an advise basis, sending exception data when measurements are outside predefined bounds,” he comments.
One of the first use cases once trending was set up on across control systems with PI was to effect nighttime setbacks of lighting and heating/ventilation services. Aside from some occupancy-controlled dimming, Genentech contol programs were designed for 24X7 consistency prior to this paradigm shift. Facilities management wanted to maintain full service 6AM-to-6PM, but then throttle back for the energy and carbon savings. Van Buskirk and his data team needed to watch the trends closely during this transition to look for manual overrides of the automation sequence and other anomalies.
“The trends within PI were a lot more complete. There are no longer 20 minute gaps in data,” he notes. “One of the analyses we ran was to run a formal data comparison of control system programming in similar buildings. You can quickly see which building is more energy efficient.”
“Centralized data provides quicker analysis and robustness of information. It is easier to determine what is needed to meet sustainability goals,” Van Buskirk summarizes. “In the future, we see potential to write more sophisticated algorithms that won’t only trigger an alarm when a situation like a high temperature is encountered, they’ll infer the ‘Why.’ For example, ‘It’s hot because someone left a door open.’ The combination of centralized and meaningful data plus smart alarming has enormous future potential for saving energy and costs.”