Príspevok
Bayesian-Inference Identification of Degradation Parameters for a Water-Cooled Variable Speed Screw Chiller
DOI: 10.18462/iir.compr.2024.0619
Stav prijatia: Abstrakt odmietnutý
Autori
Meno | Organizácia | |
---|---|---|
Andreas J. Hoess | Purdue University | |
Jiacheng Ma | Purdue University | |
James E. Braun, Ph. D. | Purdue University | |
Eckhard A. Groll, Dr.-Ing. | Purdue University | |
Davide Ziviani, Ph.D. | Purdue University |
Abstrakt
The aging and wear of mechanical components have a significant and often underestimated impact on the power consumption, efficiency, and operational safety of a vapor compression system. In this context, it is important to understand the performance degradation of screw chiller systems which have a typical lifespan of more than 15 years. Factors such as variable speed operation and oil degradation can contribute and accelerate the efficiency reduction. To investigate and quantify compressor performance degradation, a long-term experimental study was carried out on a 513 kW (146-RT) water-cooled variable speed screw chiller. The chiller underwent an accelerated life test for 4,000 operating hours. Every 1,000 operating hours, a performance test according to the AHRI 550/590 standard was conducted to assess the performance degradation behavior.
A dynamic model of the chiller was set up to simulate the chiller behavior and the compressors performance degradation. To understand the mechanisms of degradation and enable realistic prediction of performance reduction, the 4,000-h accelerated life testing and the performance tests were used to identify degradation parameters by using a Bayesian-inference approach. Based on the identified parameters, an algorithm can be developed to predict compressor performance over time and inform on predictive maintenance scheduling.
Kľúčové slová
Accelerated Life Testing,
Bayesian-inference Model,
Performance Degradation Parameters,
Predictive Maintenance,
Screw Compressor,
Chiller