Benchmark Study of CFD Prediction Accuracy of Two-Phase Flow Induced Vibration Loading for Engineering Applications

Abstract Flow induced vibration (FIV) occurs easily around the elbow of a pipe with turbulent two-phase flow such as slug flow. As a result, the FIV-induced pipe fatigue failure (PFF) may happen. It is important to prevent FIV-induced PFF from occurring when designing a piping system with potential FIV. It is considered that CFD simulation is a useful tool to predict the fluid force fluctuations or FIV loadings. The present study aims to verify CFD prediction accuracy of two-phase flow induced FIV loadings around the elbow of a pipe by comparison with the experimental data in literature. In the literature, the experiments of two-phase flow for a pipe with elbow were conducted to measure the fluid force fluctuations for various flow regimes including slug flow, churn flow and annular flow. The CFD verification study was performed for a slug flow with the same conditions as the experiment. The transient large eddy simulations (LES) of two-phase flow were carried out for 2 cases using different approaches of determining dispersed phase. The applied turbulence model is dynamic Smagorinsky LES (DLES) model. Two approaches of determining the dispersed phase are applied. One approach is that the dispersed phase is fixed. Another approach is that the dispersed phase is automatically determined by the solver based on the local volume fraction of a specific phase. CFD-predicted fluid forces acting on the pipe wall in test section were obtained in time series. The fast Fourier transform (FFT) was conducted for the time series of fluid forces to obtain the fluctuation frequency. Also, the fluid force fluctuation intensity was calculated. They are compared with the experimental data for 2 cases. CFD verification study results show that both the intensity and frequency of fluid force fluctuations predicted by DLES model and the approach of automatically determining the dispersed phase are closest to the experimental results with acceptable prediction accuracy for engineering applications.