Particle Filter System Control

This project uses novel simulation techniques to evaluate how different control strategies for central filtration systems and portable air cleaners to optimize health outcomes while reducing energy usage.

Status: Current

Research themes: Health and comfort Resilience Energy and GHG emissions

Research areas: Solutions for air leakage, ventilation and filtration; Cognitive and physical health in the built environment; Applied AI and ML Technologies for buildings; Sensors and IoT Technologies for energy and IEQ; Indoor particulate and gaseous pollutant dynamics

Project Objective

Air cleaning plays an important role in maintaining good indoor air quality. Particle filter systems (e.g., central filtration systems and portable air cleaners) are used to remove particulate matter (PM), one of the most harmful components of air pollution, from indoor air. Among other issues, these systems can require substantial amounts of energy to operate. This project examines how different control strategies can reduce system runtime to minimize energy usage while maintaining high levels of PM exposure reduction.

Approach

This project uses synthetic data derived from real measurements to evaluate the performance of different control strategies. Using this synthetic data, indoor concentrations of fine particulate matter (PM_2.5) are modelled for particle filter systems operating under different control strategies. These strategies include threshold-based control, where the system turns on above a concentrations value, and model-predictive control, where the system operates based on its predicted impact on future concentrations. Threshold-based control is evaluated using fixed thresholds and an adaptive threshold that is calculated for an individual environment.

Findings 

Threshold-based control can minimize system runtimes while maintaining high levels of exposure reduction. An adaptive threshold may help to optimize this balance. Threshold-based control and model predictive control will be compared to understand what the optimal performance of these systems is, and whether there is value in using more advanced control techniques.