Sequencing of chillers by estimating chiller power consumption using artificial neural networks

An artificial neural network (ANN) has a large memory, high capacity for learning, induction and prediction and high fault tolerance. Therefore, an ANN is ideal for developing a chiller power consumption mode that is more precise than traditional methods, since it can cope with nonlinear relationships between the power consumption, chilled water temperature and cooling water temperature of chillers in an air-conditioning system. Accordingly, this work develops an ANN-based chiller power consumption mode, to determine an optimal chiller sequencing (OCS) by way of the characteristics of the decoupled system that all online chillers have the same chilled water temperature and cooling water temperature. The test results indicate that significant savings can be made on electric rates simply by changing chiller start-up sequences.