A Performance Agent monitors assets and unit operations through KPI dashboards and self-service analytics, and a Troubleshooting Agent pinpoints the cause and recommends the fix, testable with what-if analysis before it is implemented in the plant.
TI-3145371°CHeat duty down to 1,991,631 kcal/hr from 2,396,227 earlier in the day
Reduced heat transfer efficiency, not an operational change
FoulingTube leakageVibrationPressure test the tube bundle, examine for leaks and fouling
+3.1% duty recovered · tested before implementation
KPIs and parameters are built across the plant hierarchy (plant, unit, equipment) and compared continuously against their expected norm. When a value goes off track, the alert names what moved and by how much.
Engineers explore the data themselves: correlate tags, lab results and operating parameters, compare periods, and pull trends on demand. Answers come back in the format the question needs: table, chart, correlation heat map or causal view.
Ask why the number moved. The Troubleshooting Agent reasons over the Agentic Twin™ (historical data, engineering documents and O&M records) to identify the likely cause with a confidence score, then recommends the corrective action.
Test the fix before the plant does. Scenarios run against the model (a change in feed quality, a parameter adjustment) and return the effect on throughput, quality and energy in a safe sandbox.
A complex cracking refinery needed a way to monitor and analyze KPIs and parameters of the hydrocracker unit, and to run troubleshooting tasks: information extraction, RCA and actionable recommendations on real-time equipment, process and plant performance.
Plant hierarchical KPIs with an alert system for any off-track KPI or parameter, plus a GenAI Troubleshooting Agent that answers in the format the question needs: text, tables, correlation heat maps, charts or causal analysis.
3% increase in profit across yield, production, energy and efficiency, with faster decisions at plant level or drilled down to any level of the hierarchy.
An LNG producer needed monitoring and analysis of KPIs and parameters for the acid gas removal unit of an LNG train, troubleshooting with RCA and recommendations on real-time operations, and what-if analysis.
Hierarchical KPIs with alerting, a GenAI Troubleshooting Agent for conversational analysis, and what-if analysis for real-time scenarios such as processing higher CO₂ gas while adjusting unit parameters to see the impact on throughput and the CO₂ specification to the cold box.
2.5% increase in profit across production, energy and efficiency, with scenarios tested before changes reached the train.
A batch reaction section needed real-time performance monitoring and analytics, root causes of process inefficiencies and delays, actionable recommendations to improve throughput and productivity, and visibility across process stages and operations.
KPI and parameter monitoring with deviation from benchmark, onstream sample quality prediction to cut batch cycle time, a GenAI Troubleshooting Agent wired to live KPIs for root cause analysis, recommendations and conversational operator assistance, and what-if analysis for optimum parameters and cycle time.
8% gain in production and yield with quality maintained or improved, losses reduced by 60%, people efficiency and collaboration up 25%, and ROI of 8x or better.
