Publications

Pertamina EP Cepu (PEPC) Jambaran Tiung Biru (JTB) located in Bojonegoro is a vital national gas production in Indonesia with a capacity of 192 MMSCFD experienced solvent pump several trips in 2023, disrupting production and operational stability. To address this, PEPC developed the Leading Equipment Anomaly Detection System (LEADS), an innovative solution to prevent equipment failures and minimize production losses.

LEADS leverages historical solvent pump data and employs a deep-learning LSTM Auto Encoder to model normal operational behavior. LEADS identifies anomalies by comparing real-time data to reconstructed conditions, in which when the Loss Mean Absolute Error (MAE) delta exceeds predefined thresholds it will issue early warnings to the JTB operations team for proactive response. Key features of LEADS include real-time plant visualization, anomaly trend tracking, and automated notifications, enabling seamless integration into operations.

The system categorizes anomalies into four key types: start-up (14.28%), nomination changes (42.85%), process changes (14.28%), and equipment anomalies (28.57%). LEADS has proven its accuracy in detecting anomalies on plant equipment, significantly improving early detection and preventing equipment failures as per validation performed by the JTB operations team. Besides detecting anomalies on the Solvent Pump, LEADS is also designed as a scalable platform capable of detecting anomalies in other major critical equipment throughout the plant. Implementing LEADS across major critical equipment can ensure proactive action for maintenance, reducing unplanned shutdowns and improving operational efficiency. This paper introduces a groundbreaking use of LSTM autoencoder technology for anomaly detection in gas processing plants. This innovation has been awarded the Platinum ICT Innovation Challenge also Upstream Improvement Innovation Awards from Pertamina and the 3rd Place of Digital Hackathon SKK MIGAS in 2024. It's highly scalable and has a breakthrough solution because it can be replicated and developed with expanded AI/ML technology. This solution contributes significantly to ensuring long-term operational excellence and safety.