Low Cost Indoor Air Quality Sensors

This episode is from NAFA’s Tech Seminar, recorded in April 2022. Our speaker was Dr. Jordan Clark from The Ohio State University. As in many engineering problems, the control of our indoor environments is rapidly moving from a prescriptive, rule-of-thumb paradigm to an entirely data-driven paradigm enabled by the Internet of things. The first step in such a shift is generating data reliably and cheaply with low-cost sensing.  To this end, we will look at several studies of the performance of low-cost airborne pollutant sensors, and will discuss their strengths, weaknesses, and application domain. Finally, we will look at some things we can do with these sensors once they are robust and reliable. View the slides from ⁠Dr. Jordan Clark's presentation⁠. For detailed show notes, read below and use the timestamps to navigate the episode: [2:08] Dr. Clark describes the engineering controls for air filtration in buildings. Prescribed ventilation rates and filter rates mean releasing quantities of air and waiting for feedback. We don’t have performance-based metrics or real-time feedback to avoid high costs of money and energy. An aging electrical infrastructure can’t keep up with grid peak times in buildings across all sectors and homes.  [5:07] Dr. Clark examines the evolution of prescribed air rates within Standard 62 over time from an ASHVE article. In the late 19th century, the germ theory was developed, starting the hygiene revolution that influenced the implementation of natural air ventilation. In the middle of the 20th century, central air conditioning and heating were adopted and costly. The energy crisis at the beginning of the 1980s caused the ventilation rate to decrease. Now, the ventilation rate is prescribed per person per floor area. What we should expect in the future post-pandemic is still ambiguous. One idea is to treat buildings like hospitals to reduce the transmission of viruses. We require a reevaluation of ventilation rates and consumption of energy. [9:39] Dr. Clark reviews the US Department of Energy's depiction of a grid-interactive, efficient building of the future. However, there is no mention of air quality, and that is what Dr. Clark is working on. [10:38] The first step to getting that data is using the hardware to measure quality in real-time. Air quality sensors constantly evolve, but the current focus is on low-cost particle sensors, which cost about $1,000 USD. Dr. Clark reviews the anatomy of this device. [15:15] Dr. Clark describes his performance testing as field testing with naturally occurring particle sources like gravel roads and highways. Other sources were created in the form of essential home functions like cooking. A comparison of devices and measured the sensor detection of events and if the sensor can return to baseline after an event. Other tasks included quantitating exposure to events, averaging times of performance, determining the functional range for sensors, and determining correlation and linearity among models. [23:25] Sensor performance based on sources is evaluated and found to be insensitive to building temperature. Further analysis showed that this temperature defect was a light and optical defect. Sensors are, however, significantly sensitive to humidity. Dr. Clark expands on performance as a function of size, which was evaluated to show calibration for a specific particle size. His takeaways from his research and literature help make predictions on the drift or long-term changes in performance.   [31:11] Dr. Clark concludes with gas sensors and total volatile organic compound (TVOC) detection. A total VOC sensor in low-cost particle sensor devices has limitations compared to a colorimetric Formaldehyde Sensor. The predictive abilities examined using machine learning and low-cost sensors didn't lead to any concrete conclusions.  [36:08 ] Q & A session with Dr. Clark.