Target
To predict stoppages of centrifugal compressors and identify abnormal operating modes.
Solution
The use of F5 PMM — a system for diagnosing and predicting the technical condition of equipment.
The initial data
- dataset provided: 700,000 values, with 43 parameters for each compressor
- training dataset: 150,000 values when operating in nominal mode without failures
- some of the compressor stops were known, while some were hidden by the customer
- the date and time of maintenance and repairs carried out for the analyzed period were known
Results
We constructed a mathematical model that allowed us to see abnormal operating modes and predict compressor stoppages.
The abnormal operating mode of the equipment is indicated by the growth of blue lines. This allows us to take measures to prevent incidents in advance. A red line indicates a compressor stoppage.
The abnormal operating mode of the equipment is indicated by the growth of blue lines. This allows us to take measures to prevent incidents in advance. A red line indicates a compressor stoppage.
Four compressor stoppages were predicted. These compressors were taken under special control, which, allowing the client to avoid incidents.
I express my gratitude to the team of Clover Group for their competence and professionalism. Thanks to their work, we were able to notice and prevent the shutdown of several compressors in advance. This allowed us to avoid unplanned costs and maintain our customers’ trust