Digital twins have emerged as an important tool being used in the industry for monitoring equipment. Data generated from a physical device or a process provides an opportunity to apply machine learning algorithms for anomaly detection and prediction. The insights gained by this analysis can be very useful in monitoring the physical condition of the device. The main goal of this thesis is to implement a data driven condition monitoring system for a moka pot and detect anomalies in the coffee preparation process. A data acquisition system was set up to generate data from the brewing process. A comprehensive dataset was generated which included data from ideal and anomalous iterations. Both supervised and unsupervised machine learning algorithms were trained and tested on the dataset for detecting anomalies in the process. With an accuracy score of 88%, an anomaly detection system with a reasonable performance has been implemented, demonstrating the use of the generated dataset.