This document is an expanded abstract.


A novel Plastic Bag algorithm is applied to the solution of the partial differential equations (PDE) describing the artificial gas lift (AGL) dynamics. The model is presented along with the oil and gas correlations used in the PDE model. Results of the solution, such as existence of oscillations in oil and gas flow and pressures in some operating modes, which reveals as a slugged flow, and absence of any oscillations in other operating modes, are discussed. An application of this model in a real-time simulator that is used in a hardware-in-the-loop setup is presented.


Artificial gas lift (AGL) is widely used to support and stimulate the production of oil from the reservoir when the natural flow is low. This method is favorable when the cost of gas is low or when high pressurized gas is available. The mechanism of this technology is based on dissolving the supplied gas in the oil and decreasing the hydrostatic pressure, which increases the differential pressure between the reservoir pressure and the bottom-hole pressure and results in higher oil production (Brown, 1980). A drawback of the AGL technology is the possibility of the so-called heading-casing phenomenon (instability) that is revealed as an oscillatory flow mode, which is also known as a slugged flow. These modes are hard to predict, as the dynamics of the system are quite complex. A number of publications were devoted to the development of different dynamic models of AGL.

A model having the form of partial differential equations (PDE) can be build by considering the pressure drop due to elevation change, friction and acceleration. Convenient lumped-parameter third-order dynamic models were proposed by Hu, Eikrem and Imsland (2001) and Hu (2004). It is claimed by the authors that the model allows one to reproduce the heading-casing phenomenon. An improved lumped-parameter third-order model of AGL, that accounts for the laws of the pressure distribution along the well depth in the annulus and tubing, was proposed by Hussein, Al-Durra & Boiko (2015). Also, a high-order lumped-parameter model and a PDE-based solution were proposed by Wang, Boiko & Al-Durra (2016a) and (2016b), respectively.

This content is only available via PDF.
You can access this article if you purchase or spend a download.