Success Story Process Excellence

Compressors

Compressors are integral components of many industrial facilities. Malfunctions or unavailability can lead to the failure of an entire power plant.

In the present case, problems with compressors at various points in the power plant are to be avoided, as these have repeatedly led to emergency shutdowns. In addition, energy consumption for the provision of compressed air is to be improved by detecting and diagnosing changes in the control systems and problems with leaking valves.

prewarning time before emergency shutdown

Goals

  • Early detection of unknown error types
  • Detection and elimination of process problems to improve efficiency

Challenges

  • Not all of the fault conditions that occur are always known. The greatest difficulties are caused by failures of individual components in and around compressor systems.
  • Several hundred measured variables are available for individual compressors, lines, valves and actuators. However, the components and the behavior of the components are very different.
  • Despite a large number of measurement parameters, the coverage of the sensors is limited, so there are many “blind spots”.

Solution

  • By combining pressures, valve setpoints, control inputs and other parameters, miPrexello is able to make predictions about hidden faults (combination of man and machine)
  • A malfunction of a control valve was detected at an early stage and monitored. Spare parts could be ordered in time and maintenance was be planned for a convenient time with low energy market pricies.
  • With miPrexello, a complete picture of the condition of the valves can be created, even if most of the information about them can only be derived indirectly
  • Even the smallest deviations in behavior become visible long before faults occur

Result

  • miPrexello issued a warning four days before a critical situation occurred that would have led to an emergency shutdown of the power plant
  • The actuator of a control valve had come loose. As a result, the valve was not moved to the correct position. The position itself is not measured
  • The drift between the target and predicted actual values could be closely monitored in order to plan maintenance in the best possible way.
  • In this specific case, miPrexello detected serious events that were not detected by manufacturer-specific technologies.

Your Contact

Carlos Ayala Jiménez

Carlos Ayala Jiménez

Phone:  +49 1516 2 60 74 21

eMail: