Lummus Digital enables organizations: to predict and detect process upset or unit shutdown; forecast catalyst residual life and performance, asset performance, and corrosion detection. Consequently, the life of the unit, as a whole, is enhanced.
Click and Explore our Innovative Solutions
Process Upset/ Shutdown/ Off-spec Prediction & Detection
Prediction, anomaly detection, and avoidance of process system upset/ plant shutdown, product off-spec.
Catalyst Residual Life & Performance Forecasting
To predict the residual life of a catalyst; detect anomalies due to poisoning/ variation in the plant; enhance catalyst lifetime; and avoid the reduction in reactor conversion/ unplanned shutdown for catalyst replacement.
Asset Performance (Rotating & Static)
Includes predictions, anomaly detection, cause identification, failure management/ shutdown avoidance, and machine life extension of equipment.
Asset Performance (Control Valves & Instruments)
It helps in the prediction; anomaly detection/malfunction; and cause identification of control valves and instruments’ health conditions, which include internal parts, using data science, and advanced analytics.
Corrosion Detection & Image Analytics
Includes predictions, anomaly detection, cause identification, and failure management/ shutdown avoidance due to leakage and fatigue.
Asset Performance Prediction Solution employed to Increase Uptime of an Offgas Compressor at a Major Petrochemical Group in India
- Frequent failure of Offgas Compressors leading to unplanned downtime and production loss
- Costly and time-consuming maintenance activities
- Hard to identify root cause of the problem – whether process or equipment
- Anomaly Detection: Using supervised and unsupervised machine learning, anomalies are detected along with other contributing factors that are responsible for the anomalous behavior of the equipment
- Root Cause Identification: Perform root cause analysis of the contributing factors that lead up to the equipment part/ system and generate corresponding alerts. Further, controlled variable in a data driven Failure Mode and Effect Analysis (FMEA) of equipment and application of Data Science are utilized to ascertain the same.
- Predicting Failures Ahead of Occurrence: By using deep learning Long Short-Term Memory (LSTM) of the networks, prediction of anomalies and remaining time to failure of equipment is made possible.
- Best Optimal Solution Recommendation: Conducting ‘What-if analysis’ for failure mitigation/repair management results in recommending the most appropriate solution.
Business Impacts & Outcomes
- By predicting an early compressor failure, way ahead of its occurrence, the Solution helped the Petrochemical Group steer clear of unplanned shutdown and avoid production loss amounting to an Avg. $1.2 Million per year
- Reduction in maintenance costs : Avg. $0.25 million per year
- Reduced cost of safety incidents
- Improved life of equipment as well as reduced labor and spare parts opex