In recent years, companies have executed project-specific qualification programs for subsea-processing technologies. This paper summarizes the results of a qualification program that included a multiphase, subsea-separation system for shallow-water applications. The intent of this qualification program was to develop subsea-separation technologies for the global subsea portfolio, rather than for a specific project. To meet this goal, a separator design was chosen that would meet performance targets over a wide range of operating conditions.
Introduction
Subsea processing is not a new concept; however, recent economic considerations have led to more applications, ranging from simple single-phase or multiphase boosting to separation/boosting to future compression projects. There has been a modest number of subsea-separation applications in the Norwegian North Sea, in the Gulf of Mexico, off the west coast of Angola, and most recently in the Campos basin of Brazil. Future subsea projects that have been announced include two compression and liquid-boosting units that will be installed in the Norwegian North Sea. These two projects, and a few of the installed units, use the simplest form of subsea separation: two-phase gas/liquid separation.
The most notable projects that have installed three-phase subsea separators, which remove a produced-water stream, include the Troll C pilot unit, Tordis in the North Sea, and, most recently, Marlim offshore Brazil.
Shallow-Water, Three-Phase Separator Design
In the preliminary separator design, bulk separation was provided by two inlet vane diffusers (IVDs) installed on the two inlet nozzles. In this design, the IVDs diffuse the momentum of the inlet in a gradual manner such that the liquid phases are not sheared into smaller droplets, which can lead to liquid droplets entrained in the gas or the formation of stable oil/water emulsions. Downstream of the inlet section, a series of perforated baffles was provided to straighten the flow paths in the oil/water phases in an attempt to maximize the separation length and minimize recirculation or stagnant zones. In the preliminary separator design, there were no separation internals downstream of the perforated baffles in the settling section. High-efficiency oil/water-separation internals, such as plate-pack coalescers or vessel-based electrostatic grids/coalescers, were avoided because of reliability concerns.
A water-retaining weir was included in the design to separate the oil/water outlet compartments. With this design, a single level detector, such as a nucleonic device, can be installed upstream of the weir and can be used to measure both the gas/liquid- and the oil/water-interface levels.
The majority of any entrained liquid droplets would likely be removed by gravity down the length of the separator; however, a bundle of demisting cyclones was installed in a dome at the gas outlet from the separator to perform final gas polishing.
Sand is not expected to accumulate inside of the three-phase subsea separator under favorable conditions; however, subsea well-completion failures can lead to a sudden increase in sand production. To avoid high intervention costs, dedicated sand-handling internals were included in the subsea-separator design to account for both instantaneous and prolonged sand production.
An overview of the internals configuration implemented for the shallow-water, three-phase separator is provided in Fig. 1. For a discussion of an evaluation of this design by use of computational fluid dynamics (CFD), and a validation of the CFD used, please see the complete paper.
Performance Tests on the Shallow-Water, Three-Phase-Separator Design by Use of Model Fluids
As part of the qualification program, a series of model-fluid tests was executed on the three-phase-subsea separator design by use of a low-pressure, closed flow loop (for further details about this facility, please see the complete paper). Like the CFD analyses, these tests were used to study fluid dynamics inside the separator and to optimize certain design aspects of the separator internals; however, oil and water samples were also taken during the tests to map the performance of the separator at various operating conditions.
During the low-pressure, model-fluid tests, a predefined test matrix was executed; however, additional sensitivities were also performed if unusual flow patterns, separation characteristics, or performance trends were observed. Parameters studied during the tests were flow rate, model-oil density and viscosity, and water cut. Several model oils were analyzed to determine their separation characteristics. Following a series of bottle tests, three different model oils (i.e., red diesel, naphthenic oil, and a 50:50 mixture of each) were selected.
To determine the effect of the various parameters on oil/water separation, oil and water samples were taken at the oil outlet, water outlet, and at multiple heights down the length of the separator.
A significant step change in the oil/water interface and foam/emulsion-layer thicknesses occurred around the perforated baffles in the base-case trials, particularly in the 70%-water-cut case, because of the coalescing effect of the perforated baffles. The moderate amount of shear generated by the fluid flow through the perforated baffles was effective at breaking the foam and emulsion layers formed in the inlet section. Nevertheless, stable-emulsion layers were formed because the residence time in the separator was not sufficient to break these emulsions without chemicals.
At the end of the separator, the remaining emulsion flowed over the water-retaining weir and exited with the oil outlet except in the case of the light-sensitivity trials, where the oil/water interface was controlled using the top of the emulsion layer. During these trials, it was observed that the emulsion layer was pulled into the water phase and exited through the water outlet. This effect was partly caused by the height of the vortex breaker installed on the water-outlet nozzle. As water accelerated toward the outlet nozzle, it had the tendency to pull the emulsion layer down into the vortex breaker, shearing the emulsion layer into smaller droplets and entraining oil droplets in the water outlet.
For the base-case and heavy-sensitivity trials, the oil/water interface was controlled using the bottom of the emulsion layer; the water quality was considered more important because of water-injection-quality requirements; therefore, the water quality was much improved during these trials, and the emulsion-shearing effect caused by the vortex breaker installed on the water-outlet nozzle was not an issue.
Transient flow (i.e., liquid slugging) was observed during the trials. Depending upon the given test conditions, liquid would accumulate at the base of the vertical inlet piping until the buildup of gas pressure was sufficient to sweep the line. Transient flow leads to increased shearing in the inlet piping and over the IVDs. Transient conditions are less severe at higher operating pressures for the same gas/liquid ratio; therefore, this was mostly caused by the test pressure (atmospheric) and was not considered a major concern.
Performance Tests on the Shallow-Water, Three-Phase-Separator Design With “Live” Crude Oils
As part of the qualification program, a series of high-pressure, “live”-crude tests was executed on the three-phase-subsea separator design by use of a newly commissioned high-pressure, closed-flow-loop test facility (please see the complete paper for details about this facility). Like the model-fluid tests, the tests were used to map the performance of the separator by taking fluid samples at a variety of operating conditions. These tests were inherently more representative of future field applications because of the high-pressure test conditions and use of “real” test fluids (i.e., methane, crude oil, and simulated brine).
Unlike the low-pressure, model-fluid tests, the high-pressure, “live”-crude test program did not include sand-handling trials; however, sand-removal devices (i.e., conventional sand jetting and sand-removal cyclones) were installed inside the test separator to mimic the effect of these internals on the flow patterns in the water phase.
As with the low-pressure, model-fluid tests, the test separator was equipped with a nucleonic level detector. Qualitatively, the unit performed rather well in regard to detecting the gas/liquid- and oil/water-interface levels (i.e., density bands) in the test separator; the performance was particularly better than that of the guided-wave radar that was also installed on the test separator. The nucleonic level detector was used as the primary liquid-level detector on the test separator during the high-pressure, “live”-crude performance tests.
During the high-pressure, “live”-crude tests, a predefined test matrix was executed; however, additional sensitivities were also performed if unusual separation characteristics or performance trends were observed. Parameters studied during the tests were flow rate, crude-oil density and viscosity, and water cut. Several level sensitivities were completed to further study the effect of residence time on oil/water separation. Several crude oils were analyzed to determine their separation characteristics. Following a series of bottle tests, three different crude-oil blends (i.e., “light,” “heavy,” and a 50:50 mixture of each) were selected.
To determine the effect of the various parameters on oil/water separation, oil and water samples were taken at the oil outlet and water outlet and at multiple heights down the length of the separator. Following laboratory analysis of the samples, two parameters, oil in water (OiW) and water in oil (WiO), were recorded. Online measurement of the WiO values was also provided with a WiO analyzer and a flowmeter (by density measurement) installed on the test separator’s oil outlet. Gas carry-under was measured by gas-volume-fraction meters installed on the oil and water outlets.
As would be expected, the oil/-water-separation performance of the test separator was highly dependent upon the overall flow rate through the separator. Increasing the overall flow rate at a given set of conditions means there is less chance that a given droplet will reach its respective phase before leaving the separator through either the oil or the water outlet.
Higher flow rates also lead to higher shearing of the liquids in the inlet piping and over the IVDs because of the higher velocities and inlet momentums. Generally, these shearing effects lead to the generation of smaller liquid droplets in the inlet section of the separator, resulting in more-stable, and therefore thicker, emulsion layers down the length of the separator. As more oil and water droplets are entrained in the emulsion layer or bulk fluids, the OiW and WiO values increase.
Similar effects were observed at different water cuts; the water quality generally deteriorated at high water cuts. A similar trend would have been expected for the oil quality at low water cuts; however, this effect was outweighed by the formation of stable-emulsion layers, particularly around the phase-inversion point. In the high-pressure, “live”-crude performance tests, the majority of the emulsion flowed over the water-retaining weir and exited through the oil outlet because the oil/water-interface level was controlled using the bottom of the emulsion layer. Therefore, the outlet oil quality was heavily dependent upon the stability of the formed emulsion layer. The quality of the bulk oil phase had less influence on the outlet oil quality.
Often, the oil/water-separation performance improved at higher water cuts. Emulsions formed under these conditions tend to be water-continuous with lower emulsion viscosities. Although results are not presented for the temperature-sensitivity trials, oil/water-separation performance generally improved at higher temperatures because of the lower viscosity in the oil phase and instability of the emulsion layer at higher temperatures.
Comparing the results for the different crude-oil blends at 50°C, it is apparent that the effects of oil density and viscosity led to progressively worse oil/water-separation performance from the light-sensitivity trials to the heavy-sensitivity trials.
For a discussion of the separator’s sand-handling internals and performance tests of the sand-handling equipment, please see the complete paper.
This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper OTC 25367, “Qualification of a Subsea Separator With Online Desanding Capability for Shallow-Water Applications,” by M.D. Olson, E.J. Grave, and J.C. Juarez, ExxonMobil Upstream Research, and M.R. Anderson, ExxonMobil Development, prepared for the 2014 Offshore Technology Conference, Houston, 5–8 May. The paper has not been peer reviewed. Copyright 2014 Offshore Technology Conference. Reproduced by permission.