Conference Year

2016

Keywords

Filter blockage, Indoor air flow, CART, Mann-Kendall

Abstract

This paper addresses the use of continuous indoor motor current to detect filter blockage in HVAC system. A commonly known phenomenon exists in the loading of a typical indoor motor blower that results in a power consumption decrease and hence less current draw for PSC motor, and power and currant draw increase for constant torque motor. Testing using Akaike information criterion (AIC), classification and regression tree (CART) models, and using both fixed radius basis function and linear basis function was described and performed against field of installed systems to determine if candidate data filter was sufficient, or to motivate use of Mann-Kendall to determine trend existence, strength, and transition in existence or strength. The bases, commonly used in practice, were found to have cumulative effectiveness against only 50.4% of installed systems, and were strongly differentiated in performance against motor type. The Mann-Kendall approach was found to have performance of ~88% of evaluated systems. This approach calculates the confidence trending level corresponds to the nonparametric correlation coefficient for the indoor current daily averages. Trend levels will be accumulated over time and will be used to declare filter blockage once they suggest a strong trend in the direction of filter blockage.Â

Share

COinS