Evaluating the Performance of Chiller Plant Efficiency Using Random Forest Model: A High-Rise Building Case Study

Assessing electricity consumption of chilled-water cooling plants is essential for near-optimal operation and carbon emission reduction. The goal of this study is to develop an efficient chiller sequencing control strategy for different building operating conditions. To that end, this study aims to develop three Random Forest (RF) chiller models for predicting chiller power consumption and two more efficient chiller sequencing control strategies for a 1.3 million ft² high-rise commercial office building located in New York City. Chiller cooling load, chiller power consumption, and ambient wet bulb temperature were logged at 15-min intervals in May-September 2019, and used to train RF models for analyzing the two more efficient chiller sequencing strategies. The average value of mean absolute percentage error (MAPE) and root mean squared error (RMSE) for all three RF chiller models are 5.3% and 30 kW, respectively, for the validation dataset, which confirms a good agreement between measured and predicted values. Results of this study provide additional insights on how to accurately predict the total chiller power consumption of cooling plants under different chiller sequencing control strategies.