A Two-Dimensional Mechanistic Model for Sand Erosion Prediction Including Particle Impact Characteristics

The design of oil and gas production equipment to withstand erosive conditions and optimize the production rate, while keeping the piping system operating safely, requires a reliable erosion prediction tool.  It is well known that many factors can affect erosion damage, such as flow geometry, pipe material, carrier fluid properties, flow conditions and flow regime, and particle properties.  To predict erosion, a key ingredient is to have properly calculated particle impact parameters, such as impact velocity, impact angle, impact location, and impact frequency.  The guideline in API RP 14E (1991) is not reliable in determining erosional threshold velocity when sand production is expected [Shirazi et al. 1994, Salama 1998].  A few models that were previously presented in the literature to calculate solid particle erosion utilize the fluid velocity instead of the actual particle impact velocity.  These models account for fluid density, particle diameter, and some common flow geometries, and have been compared with some lab and field data.  Their application, however, is limited due to the limited physics behind them.  Shirazi et al. (1994) presented a mechanistic model accounting for most of the key parameters listed above.  This model predicts erosion rate using the calculated representative particle impact velocity.  The drawback of this previous model is that the calculation is based on one-dimensional particle tracking.  This limits its application to relatively large sand particles (>50 to 100 microns) or cases where gas is the carrier fluid.  After extensive studies of CFD-based erosion modeling, the authors found that both the normal and tangential particle impact velocity components and the turbulence field are essential in erosion calculations for certain cases.  A mechanistic model based on two-dimensional particle impact characteristics was developed based on these findings.  Comparisons of results from the 2‑D mechanistic model and the previous 1-D model together with the CFD-based model, and experimental data are presented in this paper.