Predictive Analytics on Steam Turbine for Oil & Gas Plant
Steam turbines driving large size gas compressors in an Oil & Gas Company are among the few critical pieces of equipment. Any sudden failure would result in high-priced maintenance and colossal production loss. The data collected from condition monitoring and these machines’ process are stored on the central historian server. These are used by the plant prominent console operators to respond to abnormal alarms or by the engineering team to troubleshoot equipment following failure.
AIGC uses Data Modelling and Analytics is used to develop advanced software systems and generate intelligent decisions in real-time that improve the performance plus profitability of operations & maintenance. These applications comprehend analysis, and the data is collected into actionable information in real-time. AIGC, with the precise volume of data to the appropriate personnel minimizes any inefficiency by enabling accurate decisions at the proper time, resulting in tangible profitability. To make real-time decision support, we are also equipped with providing the most necessary data processing and modeling toolkits and heuristic modeling toolkits. Our well-trained data and domain experts can deliver large scale programs in:
- Data Acquisition – Collection of data from sources (Current & Temperature)
- Data Modeling- Includes defining the platform, configuration of the model, and running the model.
- Visual Analytics – where results in the reports are displayed with alarms.
We also offer various types of analysis, including Identification of Outliers; Performance Variation Analysis and Pattern Analysis. Our domain experts were a part of a predictive maintenance analytics program at Oil & Gas plants established to monitor steam turbines driving large-sized gas compressors. The analytics model was designed to correlate Steam Turbine (ST) operations over various loads and ambient temperatures. The solution was created, correlating all ST parameters (Steam Flow, Pressure, Temperature, RPM, Isentropic efficiency, etc.) using a broad set of historical reference data covering a full operating range. The analytics were designed to send advisories (alarms) for investigating the moment any parameter deviated from the working conditions’ predicted model.
More meticulous planning of machine maintenance
By detecting a possible machine failure, immediate solutions help plan a scheduled shutdown to repair these machines before they fail, resulting in a plant shutdown. The anomaly was detected by analytics by noticing an efficiency drop of these giant machines by about 5%. This isotropic efficiency degradation was associated with a deviation in Exhaust Steam Temperature and Exhaust Pressure.
Predictive alerts corresponding to possible breakdowns
The predictive analytics indicated a probable cause of premature seal failure and a possibility of seal contamination.
Advanced warning of impending issues
With the thrust bearing temperature trending at points higher than usual, AI solutions indicated that the gearbox shaft was slowly moving axially, thus flagging early warnings of impending issues.