Analyzing the operation of a turbine-driven feed pump
Task

To develop a model for the predictive analysis of the condition of a turbine-driven feed pump and showcase the adaptive capabilities of the Clover PMM software within the power-generating equipment maintenance and overhaul control system.

Solution

The application of Clover PMM as a system for the diagnosis and forecast of the technical condition of equipment.

Input data: telemetry data over a three-year period of operation and data regarding 10 defects with detection dates.

Sequence of actions

  • sampling was classified according to the operating conditions of the equipment and broken down into five-minute intervals to optimize the quantity of observations
  • a decision tree algorithm was selected for data processing
  • using the established dependence on the values of all variables, the algorithm calculated the value of the base variable and compared it to the actual feed water flow in the turbo pump delivery head
  • the difference between the calculated value of the feed water flow at the head of the feed turbine pump and the actual value was an error
  • an acceptable range of error was calculated for each five-minute interval. The principle of calculating the acceptable value was similar to the principle of calculating the root-mean-square deviation
  • if the estimated error fell outside the range, it was an anomaly (any „abnormal“ behavior of the equipment)

The anomalous behavior of parameters in response to an increase in feed water flow in the feed pump delivery head was registered on November 8, 2013 at 5:00 am and on November 11, 2013 at 3:30 pm. An anomalous horizontal vibration was also recorded. On November 11, 2013 the 2nd bearing of the feed pump was registered to be running at a high vibration level.
case ptn eng.png

20 groups of anomalies with high error values were detected.

Eight out of 10 represented failures were detected.
12 additional anomalous periods were detected (a combination of anomalous surges with high error values).
case ptn 2 eng.jpg
Do you want to improve the efficiency of your company?
Learn more about the possibilities of intelligent analytics and predictive analysis
* Marked fields are required
Thank you!
A Clover Group specialist will contact you during business hours.
We use cookies for your convenience on the website.
By clicking on the link to the website or using the website, you consent to the collection of data through cookies.
Learn more about our use of cookies and about changing your settings.