Sixty thousand sensors and meters collecting real-time data and information from the electrical grid, water systems, buildings and central utilities. Data fully integrated into one repository, 10 years of history with constant real-time updates of relevant information. Basic intelligence such as monitoring specifications, integrated systems data providing relationship intelligence between utilities and building performance, predictive intelligence in advance of maintenance issues and advanced analytics predicting peak power with automated power load shed. An organizational culture that mines the data, looks for improved quality and reliability for operational performance, as well as efficiency. While many view this as the art of the possible, IBM’s Vermont manufacturing enterprise has operated this way for a decade, continuously driving measurable operational performance and advanced analytics.
IBM’s Center of Excellence for Enterprise Operations in Burlington, Vermont, develops and delivers best practices for effectively running an installation of more than 30 buildings and 3.5 million sq ft of space. This secure facility includes an electrical grid that peaks at 65 MW, manages 3.2 MGD of water (water supply, industrial water production and wastewater treatment and compliance), building operations and central utility operations. The expansive integrated network of sensors monitors multiple systems, which play a critical role in achieving the quality and reliability requirements that the 24/7 manufacturing operation requires. Interconnected systems bring all the data together, which allows the team to drive continuous performance improvement with advanced data analysis and analytics. Smart grid, smart water and smart buildings are integrated to provide continuous operations to a site that hasn’t had a shutdown since 1997. In addition, the team has also driven extensive cost and environmental performance improvement. The site has an outstanding environmental and energy management record receiving at least one Vermont Governor’s Award for Environmental Excellence every year since the award’s inception 17 years ago. It has been named the Facility of the Year by Environmental Protection Magazine, is an OSHA VPP site for safety, has received multiple awards from the National Pollution Prevention Roundtable, has been recognized by the U.S. Environmental Protection Agency and Keep America Beautiful.
We will focus on the smart grid and an integrated system approach to facilities management.
The IBM Vermont smart grid includes 5,000 of the site’s 60,000 meters and sensors, which monitor parameters such as Watts, Volts, Amps, kiloWatt-hours, and sags and swells. The 65 MW system, which correlates to a small city, takes power from two 115 kV transmission lines and drives all distribution and steps down the voltage to accommodate everything ranging from 2,000-ton chillers to manufacturing equipment and office outlets. The interconnected system allows not only instant access to data, but also links to an automated paging system to provide real-time alerts. Historical and real-time data are available to not only guide system operation and maintenance actions, but are invaluable in driving the energy management program. Through data analysis and analytics the location has been able to drive a 20% reduction in energy usage and a 5 MW average peak power reduction over the past 10 years, while still growing manufacturing capability. In addition, the location has successfully participated in the ISO New England demand-response program over the past five years with an estimated $750,000 in savings over this period by voluntarily shedding electrical loads during times of peak demand on the New England grid. Through the analysis of electrical usage, the team determined which operations could be moved off peak, and which loads could be temporarily shut down under controlled conditions during peak hours.
The results are impressive, but the process starts with having a clear strategy and definition of mission critical parameters for energy management. For the IBM location, the critical objectives of the electrical grid are reliability, quality and cost. Reliability means zero downtime for a continuously operating facility, base load and peak power available all the time, and no interruptions. Only life safety systems have backup generation. Quality means semiconductor process equipment is expensive and sensitive (pure sine wave power, with voltage variation within the defined SEMI standard only). Cost encompasses a minimum 4% reduction in energy use every year. Clarity is critical. Short-term and strategic objectives need to be established for each critical parameter.
Successful strategy deployment must include an organizational design, which includes clear roles and responsibility, ownership of the measurable objectives with the data to support results and a culture of innovation that looks for new ways to do business every day. IBM Vermont’s energy management is unique in driving ownership, particularly in energy efficiency and conservation programs to drive the goals outside of the facilities organization and include mission-critical operational groups, as well as all employees. Involvement of all employees is fundamental to success as an estimated 1.5% to 2% of the energy reductions come from behavioral changes. Leadership from the top of the corporation to the top of the enterprise ensures the electrical system requirements are integrated into business operations and viewed as a key imperative, versus an extracurricular activity.
With system objectives defined and the organization established to support them, the smart grid can now be leveraged for operational performance. So how does this work?
[pullquote]Starting with reliability, let’s investigate some key areas where instrumentation leads to improved reliability and power quality. At the supply side of electricity, with a smart grid, variation of voltages can be clearly monitored in real time and instant alerts to personnel can be created. This allows immediate communication with the electric utility to identify the voltage sag, the duration of the impact (cycles) and the exact time of occurrence. This information is used to drive tactical improvements and response, but when aggregated and kept as historical information, it can be used to identify systematic issues with the transmission system. Using this approach at IBM in Vermont, the site was able to work with the utilities to improve the overall power quality to the site. One notable improvement was the utilities installation of a static synchronous compensator (statcom) device, which helped to decrease the magnitude of the voltage sags during a power disturbance. In addition, the statcom’s ability to stabilize the 115 kV transmission voltage led to a decrease in the number of operations on the site’s transformer load tap changers reducing preventive maintenance issues.
Internal electrical distribution systems are also monitored and send alarm conditions that are indicators of potential problems. Alarms can be established for an upper or lower specification, a change from a standard or a delta from a normal condition — for example, a manufacturing equipment problem presented as a transformer core and coil temperature alarm and a phase current alarm. Alarms were sent to the maintenance team’s pager. With further troubleshooting, it was determined to be a failed inverter transformer. Receiving the data and responding to the situation quickly, a significant impact to equipment was avoided. In addition proactive response reduced the cost to repair and, more importantly, did not create an unplanned impact to the manufacturing mission.
As a response to threats, assuming a reasonable amount of metering at the building level, the smart grid data allows electrical loads from each core mission to be clearly identified, with its specific power profile, versus our traditional binary understanding of whether power is on or off. Should the electrical system be at risk for a reduced supply due to area brown outs or specific targeted reductions of power supply, the data allows a disaster plan that allows for a logical shutdown of activity to respond to the power shortage while prioritizing the most critical missions. While this may seem obvious, the granularity of the smart grid data allows a mathematical comparison of load profiles within each metered area, to the available power supply. In addition, real-time demand can be monitored as load is shed from each building or the entire enterprise, to compare to the available electric supply or the supply of any other key commodity. This methodology greatly reduces the risk to operation of power impacts. At IBM, this process exists as part of a comprehensive disaster recovery plan, which outlines in detail the sequence of shedding power load should a power risk situation occur.
The energy conservation and efficiency program at IBM is sustained and delivers consistent results. While a clear energy reduction goal is imperative, the ability to monitor results of projects is an integral part of success. Many organizations complete efficiency projects with no understanding of the real impact at the meter level. Utilizing local meters, or even temporary meters, the energy performance of equipment or the infrastructure is clearly understood. As changes are made to improve efficiency, the actual results can be measured. This moves the organization from simply completing random acts of energy kindness to a set of projects that are planned and predicted for their energy usage reductions and monitoring of results.
Data, especially from across the enterprise can be a key element of generating energy usage reductions, especially when a systems approach is taken. The data not only allows for system investigation, but allows for the creation of energy productivity metrics. While the most common metric appears to be kWh/sq ft of space, more advanced analysis allows performance data such as kWh/ton of chilled water or other commodity. Take the case of the IBM Vermont chilled water system, a key commodity that is required all year to ensure conditioned air is provided to 500,000 sq ft of clean room space. From a systems perspective, there is chilled water generation, distribution and demand for cooling. Using a cross-functional team of employees including building maintenance, utility maintenance and engineering brainstorming sessions yielded multiple ideas for energy conservation (stop doing what isn’t needed) and energy efficiency (find the optimum way to do the task). What was jumpstarted in 2005 continues on today with significant results including a 19% reduction (4,200 tons) in chilled water demand and an improved energy productivity of 15% as measured in kWh/ton, saving the site $2,200 annually. And diving deeper, since outside air conditions such as temperature and dew point are drivers of the need for chilled water, the team can measure its energy productivity for those variables as well, thereby taking out the variability of weather when assessing improved performance. Add to this a culture of continuous improvement, and these highly visible metrics in the workplace create a source of challenge, pride and new ideas within the team.
Taking this to the next level, understanding the relationship of weather to chilled water, and chilled water to peak power requirements, the site can now predict its peak power based on utilizing IBM’s Deep Thunder weather model. While some may find this intellectually interesting, there is significant cost reduction opportunity. By anticipating peak power and how that relates to previous monthly peaks, enterprises can assess power load shed opportunity. This in turn allows the enterprise to avoid excessive demand charges from the utility and potentially participate in regional ISO demand response programs for payment. In addition, anticipating weather allows the utility team to anticipate how many chillers will be required for the day’s peak load, thereby allowing the team to right-size the capacity for the day, and not turn on excess equipment.
So, how does this all work? This system was started well over a decade ago and has grown and developed every year. What started as a typical SCADA system now integrates all of the various sensor and controls providers across a common communications network, integrating all data into a single repository. Efficient data storage algorithms ensure relevant data is saved, but not excessively so. Statistical applications have been established to allow that the right data to be highlighted as a serious condition, such as a pager alarm, or simply part of a daily or weekly report that requires further investigation. Add to this advanced predictive models and data analytics, and a culture of using data to measure performance and drive improvements all with a focus on adding intelligence to the system.
Instrumented, interconnected and intelligent, the IBM Vermont Center of Excellence is a proof point for the smarter base. And while the team’s success is measurable and world class, they would tell you they have only scratched the surface of what’s possible.
Janette Bombardier, P.E., is the IBM Vermont senior location executive and director of site operations. Email her at [email protected].