Optimizing demand-controlled ventilation with thermal comfort and CO2 concentrations using long short-term memory and genetic algorithm

Abstract

Heating, ventilation, and air conditioning systems are the largest consumers of energy in commercial and residential sectors. Minimizing their energy consumption without compromising indoor air quality and thermal comfort would result in environmental and financial benefits. However, most buildings still utilize constant air volume systems with on/off control to meet the thermal loads. Previous studies show that CO2-based demand-controlled ventilation is the most widely used strategy to determine the optimal level of supply air volume. Conventional CO2 mass balanced models, however, do not take the thermal comfort level into account. This paper aims to develop and implement a practical and efficient optimization framework to help office building designers and operators enhance thermal comfort and indoor air quality while minimizing energy costs for ventilation using demand-controlled ventilation. The objective function was constructed to consider thermal comfort indices, indoor air quality, and ventilation power demand. Long Short-Term Memory networks in conjunction with a generic algorithm optimization were utilized to generate the search spaces and solve the objective function. The proposed optimization strategy achieved near-optimal results within approximately 1 h of computation time. The findings indicate a reduction of the average ventilation power by 5.6% compared to the current on/off control approach and a slight increase of 0.25% in ventilation power when compared against the minimum ventilation rate recommended by ASHRAE. Additionally, the optimized approach leads to a saving of 26.9 kg per day of greenhouse gas emissions in terms of carbon dioxide equivalent.