Traditional manual chiller plant control depressed the flexibility of chiller plant optimisation and thus constituted an inefficient operation of system. Harbour City took the great initiative in retro-commissioning (RCx) for reducing power consumption and at the same time provided a comfortable indoor environment and maintained normal operation for tenants and visitors.
By reviewing the existing chiller plant control approach and the past operating data, Harbour City explored an Artificial Neural Network (ANN) and advanced Energy Modeling Technique for Chiller Plant Optimisation. Harbour City engaged in a cloud platform supplier as partner to provide technical support through Building Operating System (BOS) deployment. One year of past BMS data was extracted for machine learning and big data analysis. Chiller data profile with different parameters could also be visualised easily. After data analysis, chillers with difference in age and size were then found having different part-load performance curves. The result facilitated chiller selection and sequencing.
With the use of BOS and big data analysis, different combinations of chiller sequencing control are studied to identify the optimised COP of chiller combination with various range of cooling load at different ambient conditions. Harbour City successes in the pilot run made a practical basis for the forthcoming RCx work and much energy saving could be achieved in the near future.