Electrical Load Demand of Water Treatment Plant Profiled by Using K-Means Clustering Algorithm with Reliability Consideration

Abstract

Monitoring Electrical Load Demand (ELD) is very crucial for plants and manufacturers to observe the electrical consumption and to set the rate of a primary source, the source limit, for supplying electricity to their own electrical system. This study presents CalinskiHarabasz criterion method to find out K value in order to divide ELD of Bangkhen water treatment plant (BKP), the largest water treatment plant in Thailand and South-East Asia, as a case study into K groups. ELD of BKP recorded in twelve months is shown and analyzed by this method. Three seasons are done, too. Then, K-means clustering algorithm is applied to find the centroid value, the cluster center, of K group. The maximum cluster center is used to be the source limit to supply electricity to the plant. Moreover, this study evaluates reliability indices to find out the security of the plant from using the source limit. They are shown and discussed, when the peak load demand occurs. 