CHALLENGE
The asset was facing frequent shutdowns and the customer would like to improve its uptime. The costs to resume operations were always high, besides experiencing performance, productivity and revenue loss.
These were some struggles presented by the Head of Operations of our Energy customer when they decided to give an extra step and check if they could implement measures before having unfavourable consequences.
Their main challenge was to detect the times when the equipment was not operating as expected. These were common precursors for potentially costly errors. However, these errors were infrequent, and the failure modes were numerous.
This is a US-based company engaged in hydrocarbon exploration and production. With 60+ years of history and 4.000+ employees, the customer also operates in petroleum and natural gas gathering, processing, treating, and transportation.
SOLUTION
The customer evaluated different vendors against specific requirements and selected Kongsberg Digital as its digital twin provider based on our solid simulation experience, proven technology and the superior Hybrid Machine Learning functionality.
To emulate the plant in its normal operating conditions, physical simulators and Hybrid ML technology were implemented. Differences between the physical simulators and the real plant were used to detect potential errors by using outlier detection algorithms.
Deviations between actual and expected performance can give insight into:
- Equipment failure
- Required service intervals
- Efficiencies
- The performance is based on the following measurements:
Pressures
- Temperatures
- Flowrates
- Motor/Turbine Speeds
- Motor/Turbine Power
- Fluid Properties
IMPACT
The implementation of a real-time solution enabled the customer to detect potential errors earlier, allowing for maintenance or changes to operating conditions. Thereby, costly shutdowns or equipment deterioration could be avoided in a more efficient and smart manner.
Potential results from implementing Kognitwin Energy for Condition Monitoring:
- Detection of potential errors in advance: 30% Of alarms had an early warning flag at least 15 minutes before the alarm. Many with enough warning to do some form of intervention.
- Optimization in maintenance cycle timing and operating conditions
- Reduction of unplanned shutdowns saving 1-2 M USD/year
- Planning proper and successful turnaround activities
- Safer operations meeting government regulations
- Improved and competitive performance
- Increased team efficiency
