A first-stage three-phase separator (oil, gas, and water) on a floating oil production ship offshore (FPSO) in the North Sea was performing poorly. Upon shutting the production systems down and entering the separator, widespread damage was discovered. Engineers at AMEC Process and Energy Ltd., Aberdeen, Scotland, felt that the damage was due to forces on internal components generated by sloshing liquids under storm motion. A series of three-phase CFD analyses were used to determine the magnitude of the forces on the internal components in order to optimize the design.

Peak force values occurred at t=35 seconds, where liquid overtops the baffle

Initially, a two-dimensional problem was set up for analysis in FLUENT 5, using the VOF multiphase scheme to track interfaces for water, oil and gas. The coalescing medium was modeled as a porous region dominated by viscous losses. A number of ship pitch motions were considered, with the pitch angle varying from +/- 1° to +/- 6°. Using a time stop of 0.1 second, a transient analysis was performed in which the direction of the gravity vector changed with time to represent the pitch angle. Although this analysis did not represent heave and surge motions, it highlighted the sensitivity of the vessel to even moderate pitch angles.

Here, fluid behavior is studied at t=21 seconds in a 2D model without the baffle (above) and with the baffle (below). Sloshing was clearly suppressed when the baffle was added.

In order to suppress waves within the vessel, a baffle was conceived, consisting of a series of vertical plates face-on to the main wave direction with gaps between to admit flow. A series of two-dimensional analyses of the arrangement were performed for representative velocities. From these results, it was determined that the wave suppression baffle could be represented in the sloshing analysis by a porous region using inertial losses. Such a region was added to the 2D model and the ±6° pitching case was run again. It was clear that the sloshing had been suppressed significantly.

To study the forces on the baffle design and other internals, a set of user-defined functions was created to apply motion forces to represent first-order ship motions of surge, sway, heave, roll, pitch, and yaw as periodic functions. From actual ship motion data, the amplitudes and phase lags of points on the ship were extracted. Then, geometry and rigid mechanics were used to define the acceleration components experienced by a cell in the computational domain. Both 2D and 3D analyses were performed using the motions associated with a 50-year storm. The time step was 0.01 second and the UDF’s reported forces for the porous media representing the internal components and baffle for each time step. The results showed that peak force values occurred at 35 seconds and that at this point in the analysis, liquid is overtopping the baffle, which is properly performing its suppressive function. The analysis results provided valuable information on operating envelopes, stratified flow, sloshing, suppression and mechanical design. Local analyses were also performed to investigate alternative baffle designs. One alternative that was considered uses square hollow sections, with the faces rotated 45° to the flow, in an array. The modeling techniques enabled forces on the vessel and internals to be extracted for robust mechanical design.

An alternative baffle design was studied, which uses square hollow sections, arranged in an array, with the faces rotated 45 degrees to the flow.

FluentNEWS