Modeling and optimization of a chilled-water cooling system with multiple chillers

In order to reduce energy consumption of the centralized chilled-water cooling system in large buildings, a dynamic control strategy was proposed for cooling plants by modelling and optimization. Combined with the chilled water flow model, this paper analyzed the parallel operation characteristics of the chillers and takes the load distribution as one of the control parameters. Based on the measured data of a typical cooling system that has undergone preliminary energy-saving transformation, the residual neural network is applied to model the relationship among energy consumption, controllable parameters and environmental parameters, and the residual neural network outperforms multi-layer perceptron and support vector regression. To minimize the total energy consumption, the gray wolf optimizer was introduced to optimize the controllable variables of the cooling system. Compared with the energy consumption before optimization, the simulation energy consumption after optimization decreased 10.45% on average, while the energy saving rate is only 7.9% with equal chilled water supply temperature of parallel chillers.