Carnegie Mellon University
June 09, 2016

Achieving the Right Temperature: Making HVAC Systems More Efficient

Achieving the Right Temperature: Making HVAC Systems More Efficient In the U.S., buildings consume more than 40% of energy annually. Heating, ventilation and air conditioning systems use about half of the energy consumed by the buildings in which they operate, much of which is wasted by system faults.

To improve the performance of HVAC systems, researchers from the CEE department, including students and faculty, are working on projects with members of Facilities Management Services (FMS) – who maintain all aspects of CMU’s buildings. The data they are using comes directly from the university’s HVAC systems. Their projects, while varied, could together make these systems not only drastically more energy efficient, but also more cost effective.

Correcting Software Faults in HVAC Systems

In modern HVAC systems, software plays a major role in regulating room conditions. When a person adjusts the air settings in a room, software controls the reactions of the system to make the room hotter or colder. But if the software is not written correctly, the system may not give the person the conditions they want.

PhD student Jerry Lei is reducing errors in the software’s control logic – the part of the software that responds to commands from users – to ensure HVAC systems reach the right conditions efficiently. The control logic gathers data from hundreds of sensors in the systems that monitor everything from air temperature to air flow. If written correctly, the control logic reads the data and adjusts the system’s activities accordingly. But if there are issues with how the software reads the sensor data, the software’s control logic could wrongly adjust the system.

“It is up to the control logic to decide, based on sensor readings and other factors like outside temperature, to what extent it should activate parts of the system,” said Lei. “It is the control logic’s job to decide what … output the HVAC system produces.”

Using data harvested from on-campus HVAC systems, Lei is trying to stop faults due to errors in the control logic. If the control logic accurately tells an HVAC system how to correctly adjust a room’s conditions consistently, not only will the room be comfortable for the people in it, but the system will use less energy. An added benefit? In tandem with energy savings come cost savings.

Lei is being advised by Professors Burcu Akinci, Mario Berges, and Xuesong (Pine) Liu.

Identifying and Naming Sensors

None of Lei’s work would be possible without the hundreds of sensors within HVAC systems. The sensors are identified by names they are assigned – usually a string of letters and numbers that tell building managers what the sensor is detecting and where it is located. Currently, though, sensors across HVAC systems that detect the same measurement can be named in vastly different ways.

“Buildings automation systems maintaining HVAC systems are installed by different vendors who use different naming conventions that make it really hard to decipher” what sensors are used for, said PhD student Jingkun Gao.

Gao, advised by Professor Berges, is making it possible to automatically identify what the sensors detect by creating standard naming systems to convert often inconsistent or obscure sensor tags to more easily identifiable names. Through machine learning methods, he is assisting FMS in identifying the sensors in the campus’s many HVAC systems to fit them with uniform names.

When Gao is finished, building managers will be able to easily identify what sensors are detecting and in what buildings or rooms they are located using his standard naming systems. With this understanding, managers will be able to more quickly collect the data they need to diagnose problems.

For more information about the work the CEE department is doing to improve HVAC systems, read part two of this series, which details the work of CEE PhD student Irem Velibeyoglu as well as a collaboration between CMU and IBM.