A model-based adaptive method for evaluating the energy impact of low delta-T syndrome in complex HVAC systems using support vector regression

Abstract

Low delta-T syndrome refers to the situation where the measured differential temperature of the overall terminal air-handling units is much lower than the normal value expected. It widely exists in the existing heating, ventilating, and air-conditioning systems and results in increased energy consumption. This paper presents a model-based method to evaluate the energy impact on the chilled water pumps due to the low delta-T syndrome in a complex chilled water system. When the low delta-T syndrome occurs, the chilled water pumps would deviate from their normal working conditions with increased power consumption. Models are developed to predict the reference benchmarks of the chilled water pump power based on the current cooling load, control rules, and preset set-points. The energy impact on the chilled water pumps can be determined by comparing the measured current pump power with the predicted benchmark. Support vector regression method is introduced for predicting the chilled water flow rate of the overall terminal units. Adaptive concept is employed to enhance the prediction accuracy of the overall pressure drop of the hydraulic water network under various working conditions. The proposed method is tested and validated in a dynamic simulation platform built based on a real complex heating, ventilating, and air-conditioning system. Results show that the proposed method can accurately evaluate the impact of the low delta-T syndrome on energy consumption of the chilled water pumps. Practical application: Low delta-T syndrome widely exists in existing HVAC systems and results in increased energy consumption. This paper presents a model-based method for practical applications in assessing the energy impact on the chilled water pumps due to the low delta-T syndrome in a complex chilled water system. When the low delta-T syndrome occurs in a system, this method can be used to predict the reference benchmark of energy use of chilled water pumps based on the measured cooling load profiles, the control rules used, and the preset set-points. The energy impact can be determined by comparing the measured actual energy consumption with the predicted benchmark. The evaluation results could help the operators to conveniently monitor the energy performance of the chilled water distribution system as well as to judge whether or not taking measures to identify and correct the related faults that result in the low delta-T syndrome.