Chiller Plant 2

Chiller Plant #2

Chiller Plant inside
From 2004 to 2012, Duke University developed a Central Chilled Water System that consists of two central plants (Chiller Plant #2 is one of these) with a 42,000 ton cooling capacity and over 14 miles of distribution piping.

Themes: Distrcit Cooling, Potable water usage, energy efficiency, software, machine learning, and urbanization



Chiller Plant #2 is the one of the two major plants that make up Duke’s Central Chilled Water System. Chiller Plant #2 is responsible for cooling 74 buildings on campus and uses 15 miles of piping to do so. Chilled water leaves the plant at 38.5 degrees Fahrenheit through this network of pipes. Air handling equipment uses the chilled water to cool the buildings once the water reaches a building. The water then exits the building and returns to the plant, as part of the closed loop system, to be reused. This system operates in a variable-primary arrangement, which means that all of system’s required pumping happens at the Chiller Plant without any additional pumping equipment in buildings. There are always two people on 12-hour shifts so the system is constantly monitored.

Sustainability in Construction 

The site’s construction reflects sustainability in development. 20% of the building’s materials contain recycled content. The plant was designed with large windows to allow natural light in to save energy in this 64,000 square foot space. The red maple trees around the building were picked for their drought tolerance. Not only does Duke love the Chiller Plant, other major organizations have looked to it for inspiration during construction of their own chiller plants. Universal Studios was one such group!

Campus Functions

Keeping Campus Cool (Video)

Cooling & Humidity Control

The Central Chilled Water System (a type of district cooling system) is the most efficient and economical way to provide cooling to Duke University, Medical Center, and Schools of Medicine and Nursing. The centralized control of the pumping system decreases the amount of energy used to cool campus buildings. All energy-intensive actions are done at the plant which allows the system to be 40% more energy efficient than traditional cooling systems. Furthermore, a district cooling system increases the amount of available space on campus. Every building had to have space for a smaller chiller, a cooling tower, and air conditioner units before being part of a district cooling system. Centralizing the system’s relevant equipment allows the facilities management team to easily locate and fix equipment failures. The Chiller Plant also controls the humidity of the buildings that it is responsible for cooling. 

Water Conservation

Chiller Plant #2 was designed with sustainable water usage in mind. For example, 100,000 gallons of water are recirculated daily from the condensation on the individual Air Handling Units across campus. Rain water from the plant’s roof is collected and used as makeup water for the cooling towers. The plant draws 20 percent of its water (100 million gallons annually) from the adjacent reclamation pond that opened in 2015. The Central Chilled Water System uses 30% of all potable water consumed at Duke University. Any base water that can be supplanted by the re-use of circulated water and use of storm water drastically decreases the base amount of potable water and energy used by the chiller plants. 

Real World Relevance

District cooling systems are becoming more efficient with technological advancements. New software is able to turn formerly manually-controlled chiller systems into fully automated systems. These automated systems are able to process data from sensors throughout the system and respond in real time, without human input, to various changes within the system to ensure optimal efficiency. For example, these systems can adjust the system’s base flow to incorporate the influence of a single hallway’s electric usage. Combining this software with machine learning algorithms has already been shown to improve a system’s efficiency. Google’s AI-controlled data center has control over the very district cooling system tasked with keeping that data center from overheating. During a tornado watch at one of these data centers, the chiller plant operators noticed the system was making decisions that were counterintuitive to usual understandings of how district cooling systems function. The operators later examined the software’s decisions and discovered that the system responded perfectly to the very specific external weather conditions that were happening external to the building in order to save energy.  

Any improvement that can be made to a district cooling system’s efficiency can have monumental impact on the system’s energy use and carbon footprint since these systems generally operate on a energy usage scale of gigawatts. District cooling has not been as popular in the US as it has been in European and Middle Eastern countries. These systems increasing efficiency should help persuade more US cities to adopt district cooling systems. Using district coolers in major cities will help cities stay cool more efficiently as cities become hotter, and save space in urban areas for growing populations.

Projects at Chiller Plant #2

Project Title Project Team Project Description

Operational Performance Comparison of Variable and Fixed Speed Chillers at Duke University’s Chilled Water Plant No. 2

Myers, Andrew., Egger, Daniel., and Patino-Echeverri, Dalia.

This Master project sought to evaluate the performance of variable and fixed speed chillers in their current production environment using operational data sets from multiple chillers. The study seeked to determine the actual contribution to efficiency and conditions for efficiency of both variable and fixed speed chillers. The resulting analysis developed performance prediction models for both types of chillers under various operating conditions and in doing so demonstrates that variable speed chillers have significantly outperformed initial energy savings estimates. The results demonstrated that chiller performance and potential energy savings are relatively predictable under most operating conditions, helping to verify operators’ current understanding of plant operation.


Sverdilk, Yevgenily. (2018, August 2.) Google is Switching to a Self-Driving Data Center Management System. Data Center Knowledge.

Ferenc, Jeff. (2017, May 18.) Hershey Medical Center chiller plant optimization cuts energy use, costs. Health Facilities Management.

Pleasant, David. (2016.) US Department of Energy Recommendations and Industrial Chiller Plant Optimization.

Berbari, George. (2016, November.) District Cooling: Sustainable Design. DC Pro Engineering.

Tredinnick, Steve. (2013, June 7.)Why Is District Energy Not More Prevalent in the U.S.?. HPAC Engineering.

Cooper, Lauren & Rajkovich, Nicholas. (2012.) An Evaluation of District Energy Systems in North America: Lessons Learned from Four Heating Dominated Cities in the U.S. and Canada. American Council for an Energy-Efficient Economy. ACEEE

Waste Reduction partners. (2010.) Chillers – Energy Saving Fact Sheet. North Carolina Energy Office.