Power system diagnostics

The aim is the research and development of advanced diagnostics and reliability methods., which corresponds to the rapid development of digital signal processing technology and the development of new smart – sensors and actuators. Research and development of models and simulators as a support for technical and economic decision-making in the power systems reliability branch. New diagnostics algorithms of energy systems, especially for detection and localization of events based on time-frequency signal processing techniques associated with the physical modelling of the relevant events.

Systems for advanced diagnostics of rotary machines

Rotary machines – such as steam turbines, compressors, rotary pumps, etc. – are essential to electric power generation. Faults and subsequent outages may lead to significant losses. That is why the contemporary trend calls for on-line monitoring of such devices, providing the operator with up-to-date information on their status.

We offer modern and advanced methods of rotary machine diagnostics focusing on timely detection of imminent failures. We focus primarily on detecting and locating occurrences of rubbing, which has a significant impact on the operation of steam turbines, for example. Another important area of focus is the contactless vibration monitoring in rotating components (blades).

 

The following services are available in the area of advanced rotary machine diagnostics:

  • Measurement instrumentation, integration with existing third-party sensor and monitoring systems;
  • Adjusting in-house diagnostic HW to the needs of the customer and the application at hand;
  • SW-based assessment of equipment status (rubbing, tip-timing) relying on in-house-developed algorithms;
  • Extending diagnostic SW with additional modules specific to the given application and customer requirements.

 

References:

  • Automated rubbing detection and location system (Škoda Power, A Doosan company);
  • Rubbing detection and location in ETU TG 21 (ČEZ);
  • BTT-001 – a system for contactless blade vibration monitoring (Škoda Power, A Doosan company);
  • Processing vibration readings for cooling pumps, focusing on rubbing detection (Areva NP);
  • Developing HW (optical sensors) to measure shroud ring-fitted blade vibrations (Škoda Power, A Doosan company).

 

Design and implementation of non-stationary event monitoring systems

Technologies currently used in diagnostics and monitoring (mostly in energy industry) focus on monitoring the expected states of equipment. Under non-standard conditions, these systems cannot provide sufficient detail of the equipment condition (nuclear power plant cooling systems, gas or steam turbines, vents, compressors, etc.). Services offered below focus on such non-standard conditions.

We offer design and implementation of monitoring systems for non-standard phenomena with non-stationary characteristics – non-stationary events. Our in-house methods used in signal analysis are based on stochastic normalization of time-frequency spectrograms of the signals. Thanks to advanced evaluation methods that we use, we can provide the following event monitoring phases:

  • Detection – deciding that the event has occurred, triggering response and storing data before and after the event.
  • Location – determining the location of the event through its physical properties (for instance dispersion characteristics of tension waves propagating through the material, etc.).
  • Specification – detailed description of other quantitative properties related to non-stationary event (weight, momentum, energy, shape, etc.).

 

References:

  • Locating pulses in the EDU reactor pressure vessel (ČEZ);
  • Modernized event classifier in the KUES diagnostic system (Areva NP);
  • Detection and location of free-moving components in cooling systems (Areva NP).

 

 

Signal measurement-based machine diagnostics through time-frequency methods

Modern machine diagnostics often rely on rapidly emerging computer technologies, which have only recently enabled the development of advanced diagnostic functions. That is why in the past (and sometimes still in the present) signal processing for signals that carry the information of a potential failure in a machine or equipment focused on time analysis and a completely separate analysis on the frequency spectra.

 

We offer tools and diagnostic methods for time-frequency analysis, implemented with modern computing equipment. Original advanced algorithms of stochastic normalization have been applied to a gas turbine monitoring system; the original method of detecting tension wave dispersion curve shapes has been used to locate free-moving particles in nuclear power plant cooling systems; the method of cumulative spectrograms has been used in a system to detect steam turbine rubbing, etc.

 

We offer the following services related to time-frequency signal diagnostics:

  • Applying the newly developed methods and algorithms to specific tasks as per the client's requirements;
  • Implementing the methods using our own hardware (FPGA technology);
  • Implementing the methods using client-appointed hardware;
  • Testing under laboratory as well as real-life industrial conditions.

 

References:

  • Pulse detection in the EDU reactor pressure vessel (ČEZ);
  • System for automated detection and location of rubbing (Škoda Power, A Doosan company);
  • Modelling life-span deterioration in the blades of a high-pressure component of the TG1 ETE steam turbine (Škoda Power, A Doosan company);
  • Analysing and filtering noise in the KKU Power Plant (Areva NP).