Efficient transport of sand or cuttings is very important in the oil and gas industry, and the fluid velocity in these processes should be sufficiently high to keep particles continuously moving along the pipe. This minimum fluid velocity below which particles deposit—defined as the critical velocity—depends on various factors, including flow regime, particle size, particle concentration, phase velocities, and fluid viscosity. The objective of this study is to investigate the effect of parameters such as particle size and liquid viscosity on solid/particle transport in horizontal pipelines by use of computational-fluid-dynamics (CFD) simulations and to validate the numerical-model predictions with experimental data. Also, a mechanistic model that is based on force balance is proposed to predict the critical velocity under various experimental conditions.
CFD simulations have been conducted with a commercially available software (ANSYS-FLUENT). An Eulerian model with a k-w shear-stress transport (SST) turbulence-closure model is used to simulate the fluid flow while particles are tracked as the Lagrangian phase. In these simulations, an eddy-interaction model is included to consider the effect of flow turbulence on particle tracking. The simulations are created for a 0.05-m pipe diameter with a 4-m length. The simulations are initialized at relatively high fluid velocity, which is gradually reduced until the particle velocity drops below the acceptable critical velocity.
The CFD simulation and proposed mechanistic model results are validated with experimental data from literature (Najmi 2015; Najmi et al. 2016) for two particle sizes and multiple liquid viscosities. The simulation and model results show that, depending on the flow regimes (laminar or turbulent) and particle size, the critical velocity demonstrates a similar trend with carrier liquid viscosity as that of the experimental data. However, both the CFD and developed models show poor performance for higher particle size (600 µm). Also, the CFD simulations, experimental data, and proposed-model results are compared with three models currently used in the industry, namely, the Oroskar and Turian (1980) model, the Salama (2000) model, and the Danielson (2007) model.