Computational Fluid Dynamics (CFD) are used to simulate flow-induced vibrations (FIV) in high-pressure multiphase pipe flow. Furthermore, empirical correlations from the literature is compared and validated against computational and experimental results. Based on the CFD results and in conjunction with the reference 6" (internal diameter (ID)) data new scaling rules are proposed.


Flow-Induced Vibration (FIV) in subsea production systems (SPS) may lead to fatigue fracture in piping, support structure and at welds. There are multiple physical mechanisms through which internal flow in subsea production systems may induce a vibrating response of the structure and fatigue issues. One of these mechanisms involve multiphase flow induced excitations, where variations in density lead to time-varying reaction forces that induce time-varying motion of the piping. Currently used empirical relations are mainly based on low pressure experiments. These relations provide a description of the frequency spectrum of the dynamic loads induced by multiphase flows and are partly validated with field case observations and measurements. Despite this, former investigations have shown that the remaining uncertainty is still large, for instance regarding the width of the frequency spectra at high pressures. In this paper, Computational Fluid Dynamics (CFD) simulations are used to understand impact of certain parameters and to bridge the gaps between the low-pressure experiments vs. field specific conditions. The CFD methodology is validated against literature data. With the same schemes, field cases are simulated. Based on the CFD results, adapted scaling rules for the force spectrum are derived and compared against the available laboratory data. The mechanical response due to multiphase flow can be calculated either in the time domain or in the frequency domain. In the time domain, the wave/slug frequency and heights are required including the wave/slug velocity. This is often the most accurate description and can for instance be obtained from CFD. On the other hand, the mechanical simulations are often easier in the frequency domain. From previous experiments, it turns out that the power spectral density (PSD) of the force looks like a "triangle" in the log-log scale (Figure 1). This means that to describe this PSD for general conditions, four relations are required: The peak frequency, the slope parameters m1 and m2 and the total rms value (equals the square root of the integral of the PSD). Therefore, correlations for these four parameters are sought-after for a fast prediction of the expected forces.

This content is only available via PDF.
You can access this article if you purchase or spend a download.