Discrete model-based operation of cooling tower based on statistical analysis

This study is aimed to utilize the operation data to build a physical-meaningful and precise-enough model to assist the operation of a cooling tower. To do so, this work introduces a dimensionless index, which can describe the cooling capability of a cooling tower in terms of effective power utilization. In the first phase of this study, principal component analysis, one of factor analysis methods, is used to investigate effects of ambient air temperature and relative humidity on the cooling capability of a cooling tower. Based on the proposed cooling capability index, the operation data are partitioned into different groups by the fuzzy c-mean clustering algorithm. The resulted groups are distinctly categorized by the conditions of ambient air temperature and relative humidity. In the second phase of the study, data within the same mode of a set of fans are partitioned by the fuzzy c-mean clustering algorithm. The resulted groups of data are then modeled by linear regression. The acquired multiple models are highly accurate in predicting the output temperature of cooling water from the cooling tower. The acquired models assist the operator to accurately select the proper fan mode when process conditions, e.g., cooling loading, or environment conditions, e.g., ambient air temperature, change. It results in electricity saving. This study is concluded by the presentation of a discrete model-based approach to determine the fan mode. The application to a real cooling tower in an iron and steel plant is promising in saving electricity consumed by the fan set.