An algorithmic approach to predicting mechanical draft cooling tower fan speeds from infrasound signals. (October 2022)

• Record Type: Journal Article
• Authors: Eaton, Samuel W.; Cárdenas, Edna S.; Hix, Jay D.; Johnson, James T.; Watson, Scott M.; Chichester, David L.; Garcés, Milton A.; Magaña-Zook, Steven A.; Maceira, Monica; Marcillo, Omar E.; Chai, Chengping; d'Entremont, Brian P.; Reichardt, Thomas A.
• Is Part Of: Applied acoustics. Volume 199(2022)
• Publication Date: 2022-10
• Subjects: Mechanical draft cooling tower -- Blade passing frequency -- Fan speed -- Nuclear research reactor; Acoustical engineering -- Periodicals; Periodicals; 620.2
• DOI: 10.1016/j.apacoust.2022.109015
• Languages: English
• ISSNs: 0003-682X

Abstract

Highlights: This advance provides a new tool for cooling tower performance monitoring and maintenance. Infrasound data was collected persistently outside of two nuclear research reactor cooling towers. An algorithm was developed to extract cooling tower fan speeds from that infrasound data. The approach was found to be transferable between facilities as well as between microphone types. Abstract: Mechanical draft cooling towers (MDCTs) serve a critical heat management role in a variety of industries. For nuclear reactors in particular, the consistent, predictable operation of MDCTs is required to avoid damage to infrastructure and reduce the potential for catastrophic failure. Accurate, reliable measurement of MDCT fan speed is therefore an important maintenance and safety requirement. To that end, we have developed an algorithm for automatically predicting the rotational speeds of multiple, simultaneously operating fan rotors using contactless, infrasound measurements. The algorithm is based on identifying the blade passing frequencies (BPFs), their harmonics, as well as the motor frequencies (MFs) for each fan in operation. Using the algorithm, these frequencies can be automatically identified in the acoustic waveform's short-time Fourier transform spectrogram. Attribution is aided by a set of filters that rely on the unique spectral and temporal characteristics of fan operation, as well as the intrinsic frequency ratios of the BPF harmonics and the BPF/MF signals. The algorithm was tested against infrasound data acquired from infrasound sensors deployed at two research reactors: the Advanced Test Reactor (ATR) located at Idaho National Laboratory (INL) and the High Flux Isotope Reactor (HFIR) located at Oak Ridge National Laboratory (ORNL). After manually identifying the MDCT gearbox ratio, the algorithm was able to quickly yield fan speeds at both reactors in good agreement with ground truth. Ultimately, this work demonstrates the ease by which MDCT fans may be monitored in order to optimize operational conditions and avoid infrastructure damage.