Assessing operating statuses for chiller system with Cox regression

The coefficient of performance (COP) of chiller systems drops when their operating statuses change frequently and deviate from the conventional pair-up operation of chillers, pumps and cooling towers. This paper analyses the chance of having transitional and unconventional operating statuses when the operating variables vary. 12,541 sets of operating conditions were collected from a chiller system at 15 min intervals over 8 months. 1901 sets of transitional statuses and 1565 sets of unconventional statuses were identified. Cox regression models were developed to analyze the impacts of ambient variables, flow- and temperature-related operating variables on transitional statuses. The temperatures of water entering and leaving the condensers were significant variables influencing the chance of switching on/off an extra cooling tower. Survival curves suggested the throttling range of significant variables to identify the chance of transitional statuses causing a COP fluctuation. Minimizing unnecessary transitional and unconventional statuses increased the average system COP by 2.72%.