In practice, the energy consumption of industrial equipment rises mostly due to wear and tear, which might include leaks or faulty plant conditions. By using statistical techniques and artificial intelligence, energy managers can promptly identify abnormal energy usage and determine the underlying causes for such anomalies. There are various machine learning algorithms powerful enough to become anomaly detectors; however, relevant candidates must be sufficiently accurate and rapid response for industrial application. Thanks to suitable performance on different types of unsupervised anomaly detection, the Local Outlier Factor approach has been selected. The aim of this study is to identify abnormal energy consumption patterns for preventing energy loss and promoting predictive maintenance in a beverage processing factory in Vietnam. The anomaly score has been evaluated to detect multivariate and subsequence anomalies. Finally, energy consumption time-series with anomalous points highlighted has been provided for the business user to be actionable on the anomaly.