Integrated probabilistic modeling method for transient opening height prediction of check valves in oil-gas multiphase pumps. (April 2018)

• Record Type: Journal Article
• Authors: Deng, Hongying; Liu, Yi; Li, Ping; Ma, Yi; Zhang, Shengchang
• Is Part Of: Advances in engineering software. Volume 118(2018)
• Page Start: 18
• Page End: 26
• Publication Date: 2018-04
• Subjects: Check valve -- Multiphase pump -- Computational fluid dynamics -- Probabilistic model -- Gaussian process regression; CFD Computational fluid dynamics -- GPR Gaussian process regression -- IPM Integrated probabilistic modeling -- MARE Maximum absolute relative error -- MECP Mean ensemble conditional probability -- MP Maximum MECP -- RE Relative root-mean-square error; Computer-aided engineering -- Periodicals; Engineering -- Computer programs -- Periodicals; Engineering -- Software -- Periodicals; Periodicals; 620.0028553
• DOI: 10.1016/j.advengsoft.2018.01.003
• Languages: English
• ISSNs: 0965-9978

Abstract

Highlights: An integrated probabilistic modeling method is developed to predict the transient opening height of the check valve in oil-gas multiphase pumps. Both advantages of computational fluid dynamics (CFD) and probabilistic models are integrated. A suitable CFD turbulence model is selected efficiently by evaluation of the reliability of candidate models. Simulation and experimental results show the feasibility and simplicity of the proposed method. Abstract: To ensure the reliability of oil-gas multiphase pumps, it is necessary to model the relationship between the transient opening height of the check valve and multiphase transportation conditions. However, the choice of a suitable turbulence model is not straightforward for the widely used computational fluid dynamics (CFD) method. Additionally, the CFD models are still not easy to be validated experimentally because the transient opening motion is difficult to be accurately measured for its quick and nonlinear characteristics. In this work, an integrated probabilistic modeling (IPM) method is proposed to predict the transient opening height of the check valve in oil-gas multiphase pumps. First, candidate CFD transient models with different turbulence models are adopted to provide initial training data for several Gaussian process regression (GPR) models. Then, a probabilistic index of the trained GPR models is formulated to assess their reliability. Consequently, without knowing the actual values, a suitable GPR model is chosen from the candidates for online prediction of a new condition. Moreover, instead of the time-consuming CFD design process, a better CFD turbulence model can be selected efficiently. The superiority of IPM is demonstrated using simulation and experiments.