Fluid dynamic study of cyclonic separators to oil wells through CFD techniques

Sand production control in oil and gas wells has been a subject of great interest in the oil industry. Sand production has a direct implication in increasing the costs of production and has been widely described in the literature. Several authors have reported problems due to the sand production in wells that includes: premature waste of equipment, a considerable increase in the number of workover operations, and the availability of rigs to attend the demand of workover operations. In this context, this paper presents a study focused on the numerical simulation of fluid dynamics inside a downhole desander device cyclonic separator considering 32 parameters modifications. The proposed designs varieties are the following: change in five geometrical parameters; one boundary condition and two-fluid properties. These changes will allow the application of fractional factorial design with two levels to evaluate the effect of these parameters (inner pipe external diameter, inner inflow tubing diameter, number of helical channels, helical channel slops, vortex finder length, volumetric flow rate, viscosity, and density) over tangential velocity and pressure drop. To achieve this goal, a three-dimensional, single-phase, and isothermal model was implemented in a commercial CFD code. Laminar and turbulent regimes were considered. According to the results, it was verified that for the five geometrical factors analyzed the inner pipe external diameter, the number of helical channels and the vortex finder length have the biggest impact on the tangential velocity presenting an increase of 7.95 m/s, which represents an increase from 60 m^3/d to 116 m^3/d in the volumetric flow rate capacity. On the other hand, the inner pipe external diameter, the flow rate and the number of helical channels are the parameters that most influences the pressure drop, among the results, an increase from 2 3/8'' to 3 1/2'' in the inner pipe external diameter yields a reduction of 0.5 MPa. Finally, a regression model was proposed for the pressure drop as a function of the eight parameters analyzed and the main interactions between two factors.