Reducing False Alarms for Early Kick and Loss Detection (EKLD) in Real-Time Application Using Knowledge-Assisted Case-Based Reasoning

Abstract Recent advancements in Early Kick and Loss Detection (EKLD) have leveraged various sensor information in real-time for detection and time-series trend analysis. However, the increase in data and complexity has led to a higher rate of false alarms, undermining the reliability and usability of the system. In this study, we developed a Knowledge-Assisted Case-Based Reasoning (KACBR) approach by integrating historical data with contextual knowledge to reduce the false alarm rate while maintaining detection capabilities. The expert system employs a hybrid methodology that combines the structured domain knowledge from Knowledge-Based Expert Systems (KBES) with the flexibility, robustness, and adaptability of Case-Based Reasoning (CBR). The model was validated and evaluated using real-time drilling data from the field. Employing the hybrid approach developed in this study, significantly reduced the overall false alarm rate while maintaining high performance in event detection. Field testing in real-time confirmed the system's effectiveness in detecting both kick and lost circulation events across all drilling phases without compromising the speed of real-time monitoring. Comparative analysis with traditional kick and loss detection methods and statistical trend-analysis-based systems revealed that the Knowledge-Assisted Case-Based Reasoning (KACBR) approach excels in reducing false alarms and maintaining detection reliability.