Energy-efficient operation of portable air cleaners based on real-time prediction of non-uniform concentrations of indoor air pollutants in open offices

Abstract

In an open office environment with high occupant density and random occupancy patterns, excessive energy consumption is often observed because mechanical ventilation (MV) is usually designed and operated assuming a full-occupancy-based ventilation rate (VR). Considering the duo challenges of post-pandemic and climate change, this study proposes a real-time monitoring and optimization approach for operating portable air cleaners (PACs) to assist and reduce the energy consumption of the MV while maintaining the minimal VR and improving indoor air quality (IAQ). The approach was as follows. First, by assuming four VRs and 36 different emission sources, numerical simulations were conducted and validated based on an actual open office on non-uniform concentrations of a surrogate for indoor air pollutants (IAPs) throughout the breathing zone. Second, the pre- and post-purified IAP distributions were obtained using a limited number of sensors and trained by two artificial neural networks. Third, the real-time optimization of PACs’ on/off operation and placements was accomplished by the particle swarm optimization algorithm to balance energy output and improve IAQ simultaneously. Results showed that by deploying four sensors, the predictions of post-purifying concentrations were in acceptable accuracy (i.e., CV-RMSE <2.0%) within 30–40s. Using at most three PACs resulted in a significant decline in local IAP concentration levels. Meanwhile, an average reduction of total energy consumption of MV and PACs was 34.6% compared to using MV to reach the same levels. Overall, this study supports the use of PACs to assist MV in achieving energy efficiency and good IAQ.