기계학습을 이용한 유동가속부식 모델링: 랜덤 포레스트와 비선형 회귀분석과의 비교
Modeling of Flow-Accelerated Corrosion using Machine Learning: Comparison between Random Forest and Non-linear Regression
0
조회
0
다운로드
3
인용
0
참고문헌
키워드
초록
Flow-Accelerated Corrosion (FAC) is a phenomenon in which a protective coating on a metal surfaceis dissolved by a flow of fluid in a metal pipe, leading to continuous wall-thinning. Recently, manycountries have developed computer codes to manage FAC in power plants, and the FAC predictionmodel in these computer codes plays an important role in predictive performance. Herein, the FACprediction model was developed by applying a machine learning method and the conventional nonlinearregression method. The random forest, a widely used machine learning technique in predictive modelingled to easy calculation of FAC tendency for five input variables: flow rate, temperature, pH, Cr content,and dissolved oxygen concentration. However, the model showed significant errors in some inputconditions, and it was difficult to obtain proper regression results without using additional data points. In contrast, nonlinear regression analysis predicted robust estimation even with relatively insufficientdata by assuming an empirical equation and the model showed better predictive power when the interactionbetween DO and pH was considered. The comparative analysis of this study is believed to provideimportant insights for developing a more sophisticated FAC prediction model.
Abstract
Flow-Accelerated Corrosion (FAC) is a phenomenon in which a protective coating on a metal surfaceis dissolved by a flow of fluid in a metal pipe, leading to continuous wall-thinning. Recently, manycountries have developed computer codes to manage FAC in power plants, and the FAC predictionmodel in these computer codes plays an important role in predictive performance. Herein, the FACprediction model was developed by applying a machine learning method and the conventional nonlinearregression method. The random forest, a widely used machine learning technique in predictive modelingled to easy calculation of FAC tendency for five input variables: flow rate, temperature, pH, Cr content,and dissolved oxygen concentration. However, the model showed significant errors in some inputconditions, and it was difficult to obtain proper regression results without using additional data points. In contrast, nonlinear regression analysis predicted robust estimation even with relatively insufficientdata by assuming an empirical equation and the model showed better predictive power when the interactionbetween DO and pH was considered. The comparative analysis of this study is believed to provideimportant insights for developing a more sophisticated FAC prediction model.
참고문헌 (15)
- Flow-Accelerated Corrosion, "B. Chexal", EPRI, ~, pp.EPRI TR-106611-R1 (1998)
- Assessment and avoidance of erosion-corrosion damage in PWR feedpipework, "I. S. Woosey", IAEA, 60~65, pp.IAEA Report:IWG-RRPC-88-1 (1990)
- G. J. Bignold, "Erosion-corrosion in nuclear steam generators", pp.5~18 (1981), Water Chemistry of Nuclear Reactor Systems
- J. Ducreux, "Theoretical and Experimental Investigation of the Effect of Chemical Composition of Steels on their Erosion-Corrosion Resistance", Specialists Meeting on Erosion-Corrosion of Steels in High Temperature Water and Wet Steam (1982)
- "EPRI TR-103198-P1, CHECWORKS Computer Program Users Guide", EPRI, ~ (1997)
- W. Kastner, , Nucl. Eng. Des, 119, pp.431~ (1990)
- S. Trevin, "Making FAC Calculations With BRT-CICERO™ and Updating to Version 3.0", 17th International Conference on Nuclear Engineering, 1, pp.81~88 (2009), Plant Operations, Maintenance, Engineering, Modifications and Life Cycle; Component Reliability and Materials Issues; Next Generation Systems, Brussels, Belgium
- 황경모, "CHECWORKS와 ToSPACE 프로그램의 배관감육 해석결과 비교", CORROSION SCIENCE AND TECHNOLOGY, 17(6), pp.317~323 (2018), 한국부식방식학회[DOI]
- M. Kuhn, "Applied Predictive modeling" (2013), Springer
- 이경근, "유동가속부식에 대한 통계적 모델링 해석방법 개발: 유속, 온도, pH 및 Cr 함량의 효과", 한국압력기기공학회 논문집, 12(2), pp.40~49 (2016), 한국압력기기공학회[DOI]
- Effects of Water Chemistry and Fluid Dynamics on Wall Thinning Behavior <Part 2> - Evaluation of Influential Factors by Flow Accelerated Corrosion Tests, "K. Fujiwara", CRIEPI, ~, pp.CRIEPI-report: Q09028 (2009)
- Effects of Water Chemistry on Wall Thinning Behavior in Single - Phase and Two -Phase Flow, "K. Fujiwara", CRIEPI, ~, pp.CRIEPI-report: Q11025 (2012)
- "A Language and Environment for Statistical Computing, R Foundation for Statistical Computing" (2008)[URL]
- "Package ‘randomForest’" (2018)[URL]
- "Package ‘minpack.lm’" (2016)[URL]