Fault Detection Method for Residential HVAC Systems

From ASHRAE Journal Newsletter, May 26, 2020

Many of the methods for fault detection and diagnosis (FDD) in air conditioning systems are developed for packaged commercial systems.

A recent Science and Technology for the Built Environment article presents a way to obtain key operating parameters for air conditioning systems (data such as airflow rate, cooling capacity, system efficiency, and refrigerant mass flow) so the information can be used for air conditioning systems in the residential sector.

Researchers Austin P. Rogers, Ph.D., Associate Member ASHRAE, a mechanical engineer at Pacific Northwest National Laboratory; Fangzhou Guo; and Bryan Rasmussen, Ph.D., Member ASHRAE, discuss the research.

1. What is the significance of this research?

Our work investigated and proposed a method, designed specifically for residential air conditioning systems, for detecting and diagnosing faults. The residential air conditioning market poses unique challenges for commercializing technologies such as fault detection and diagnostics (FDD). For example, the cost of extra sensors required for FDD is harder to justify in these smaller units, and having two separate packages (i.e., an indoor and an outdoor unit) complicates sensor installation.

We recognized that the state-of-the-art research, while effective for larger commercial systems, is not appropriate for the residential market, and we proposed compromises in diagnostic capability in order to reduce cost and simplify installation. Specifically, we used a reduced number of sensors installed only in the indoor unit to detect faulty operation and distinguish between air-side and refrigerant-side faults.

2. Why is it important to explore this topic now?

Commercial buildings have traditionally been a focus for advanced heating, ventilation and air conditioning (HVAC) controls, such as providing grid services and FDD. However, as our energy and electricity outlook transforms, the residential sector will need to be transformed as well.

We need to ensure efficient operation of high-energy-consuming residential systems such as HVAC, but solutions in the residential market are different than their commercial counterparts. Initial costs need to be lower in the residential market, and this requires a prioritization of features, with only the most important features included.

3. What lessons, facts, and/or guidance can an engineer working in the field take away from this research?

Of course, we hope that this work will guide engineers incorporating FDD in their products. More generally, this work highlights an important tool for analyzing energy systems; certain modes of operation provide more direct insight to certain performance characteristics. In our work, we found that the airflow delivered to the home is easier to estimate while heating instead of cooling, but this principle will apply in different ways to many systems.

4. How can this research further the industry's knowledge on this topic?

This research was a result of engineers in the industry wanting to provide value to their customers at an attractive price. The research teaches us that, if we are willing to limit the number of faults that we are able to diagnose, then we can detect faults and perform some diagnostics using only a few sensors. We focused on distinguishing between airflow faults (e.g., fouled air filter, crushed ductwork, closed supply registers) and refrigerant-side faults (e.g., incorrect refrigerant charge, flow restriction). This provides valuable insight without the full cost of a more sophisticated FDD method.

5. Were there any surprises or unforeseen challenges for you when preparing this research?

Over the past 10 years, “virtual sensors” have been used to make the air conditioning FDD process simpler and more intuitive. We needed to understand the assumptions and actual sensors needed to create a virtual sensor measurement, and then we needed to compare and contrast different virtual sensing methods. In order to address this challenge, we represented different methods in a block diagram format in order to highlight the estimation process and the assumptions involved.