An Improved Model for Predicting Separation Efficiency of a Rotary Gas Separator in ESP Systems
- A.F. Harun (INTEC Engineering) | M.G. Prado (U. of Tulsa) | S.A. Shirazi (U. of Tulsa) | D.R. Doty (U. of Tulsa)
- Document ID
- Society of Petroleum Engineers
- SPE Production & Facilities
- Publication Date
- May 2002
- Document Type
- Journal Paper
- 78 - 83
- 2002. Society of Petroleum Engineers
- 3.1.2 Electric Submersible Pumps, 3.1.1 Beam and related pumping techniques, 4.3 Flow Assurance, 5.2.1 Phase Behavior and PVT Measurements, 5.2.2 Fluid Modeling, Equations of State, 4.1.5 Processing Equipment, 4.11.1 Corrosion Research, 3.1 Artificial Lift Systems, 4.2.3 Materials and Corrosion, 4.1.2 Separation and Treating, 5.3.2 Multiphase Flow
- 0 in the last 30 days
- 567 since 2007
- Show more detail
- View rights & permissions
|SPE Member Price:||USD 5.00|
|SPE Non-Member Price:||USD 35.00|
An improved model capable of predicting the separation efficiency of a rotary gas separator (RGS) in electric submersible pump (ESP) systems is presented. The model incorporates a new, twophase, flow-inducer model capable of calculating the inducer head. The inducer head, generated by an RGS, has been identified as a key parameter that distinguishes between a separator's high- and low-efficiency regions. This information was previously determined empirically but can now be calculated. The new model more accurately predicts the maximum liquid rate at which an RGS should be installed. A comparison of the model's predictions with water/air and hydrocarbon/air experimental data indicates that the improved model performs better than earlier ones.
In designing ESP systems for gassy oil wells, an RGS is still one of the most commonly used gas-handling devices, capable of minimizing the amount of free gas going into the ESP pump section.
An ESP system design for gassy oil wells should address two important questions.
How much free gas can an RGS separate?
At what maximum liquid rate, if any, should the RGS be installed?
In an attempt to answer the previous questions, Alhanati1 developed a mechanistic model to predict the separation efficiency of an RGS. A typical separation-efficiency curve, as predicted by the Alhanati1 model, shows three distinct efficiency regions (high, transition, and low) as a function of the liquid flow rate (Fig. 1). The transition region corresponds to the flow rates in which the inducer is overloaded and is no longer able to provide enough head to compensate for the pressure drop across the outlet ports. Because of a lack of available information, the Alhanati1 model relies on an empirical inducer-head curve that was developed indirectly from separation-efficiency data (Fig. 2). Because the inducer-head curve ultimately dictates the exact location of the efficiency transition region, Alhanati1 suggests that one should avoid designing an RGS installation close to this region. In addition, his model must be used with extreme caution when applied to a different RGS unless the inducer-head curve for that particular RGS is provided as additional input into his model.
The objective of this study is to improve the Alhanati1 model by developing an inducer model that calculates the inducer head directly, based on both the inducer geometrical data and the system operating conditions. This paper briefly explains how the inducer model is developed and incorporated into the Alhanati1 model.
The separation process in an RGS installation occurs primarily within two distinct flow domains - the tubing-casing annulus and the RGS centrifuge chamber. The natural separation process within the annulus affects the actual amount of gas and liquid going into the separator. The separation process inside the centrifuge chamber affects the amount of gas and liquid being expelled back into the annulus.
The link between these two separate flow domains depends upon the separator characteristics (i.e., the head generated by the inducer and the pressure drop across the gas outlet port). The inducer has to generate sufficient head to compensate for the pressure drop across the gas outlet port so that the separated gas can be expelled back into the tubing-casing annulus.
Based on his model, Alhanati1 discovered three efficiency regions-high, transition, and low (Fig. 1). The high-efficiency region reflects the separation efficiency of both natural separation within the tubing-casing annulus and separation caused by centrifugation inside the separator. The transition region is thought to occur when the inducer is no longer able to generate enough head to compensate for the pressure drop across the outlet port. The low-efficiency region reflects only the natural separation efficiency occurring within the tubing-casing annulus section because the separation process has ceased.
Because of a lack of information, Alhanati1 developed an empirical inducer-head curve that "best matches" his specific experimental data (Fig. 2). Based on the fact that the inducer is considered as a low net-positive suction head (NPSH) impeller, Alhanati1 and Lackner2 assumed that the inducer-head curve is not affected by the presence of free gas. However, Soltis et al.3 showed that inducer-head degradation is possible when free gas is introduced. Therefore, the previous assumption needs to be investigated further. To investigate the effect of gas on inducer performance, Harun4 developed the following two-phase, mechanistic flow model to predict the generated inducer head as a function of geometry, fluid properties, and operating conditions. The inducer model was then incorporated into the Alhanati1 model so that the location of the transition region could be predicted better.
Inducers are classified as axial flow pumps and are characterized by helical path blades (Fig. 3). As reported by Ross and Banerian,5 Acosta,6 Rains,7 and Lakshminarayana,8-10 inducers have been used in the aerospace industry to increase the specific suction speed of liquid rocket pumps so that cavitation problems can be avoided.
Because two-phase flow modeling in turbomachinery is a complex subject, the current modeling approach is based on a simple, single-phase model, which is then expanded to handle two-phase flow. For that reason, the meridional-flow solution technique- originated by Hamrick et al.,11 modified by Katsanis,12 and applied specifically to inducer blades by Cooper and Bosch13-has been used as the starting point for this model. The single-phase model formulation is expanded into a two-phase model with the two-fluid modeling approach. The void fraction equation is added to the set of governing equations to calculate the local void fraction, thereby allowing slippage between the phases to occur. The interfacial friction coefficient becomes a required closure relationship in the void fraction equation.
|File Size||1 MB||Number of Pages||6|