Abstract

The gas well loading phenomenon is considered one of the most serious problems in the gas industry. It occurs as a result of liquid accumulation in the wellbore, when the gas phase does not provide sufficient energy to lift the produced fluids, imposing additional hydrostatic pressure on the reservoir and causes more reduction in the transport energy. Eventually, if the reservoir pressure is low, then the accumulated liquids may completely kill the well. If the reservoir pressure is higher, then liquid slugging or churning may take place and gives more chance for liquid accumulation and the well may die. To solve such a problem, the well may be unloaded mechanically using a pump or by gas lift; or to let the well continuously unloading itself. Analysis of the mechanisms of gas-well load-up indicates that there is a critical gas rate to keep low pressure gas wells unloaded. Predicting that minimum gas flow rate is very crucial. Several researchers have developed various mathematical models to calculate the critical flow rate necessary to keep gas wells unloaded.

This paper presents an Artificial Neural Network (ANN) model for predicting the minimum flow rate for continuous removal of liquids from the wellbore. The model is developed using field data from different gas wells. These data were used to train a three-layer backpropagation neural network model. The model was tested against published field data which was not used in the training phase. The results show that the developed model provides better predictions and higher accuracy than the published models. The present model provides prediction of the critical gas flow rates with an absolute average percent error of 4.61 % and a correlation coefficient of 99.11%.

Introduction

The problem of gas well loading is serious one in gas wells. Due to lack of sufficient energy to lift the produced fluids, liquid may accumulate in the wellbore, and impose additional hydrostatic pressure on the reservoir and causes more reduction in the transport energy. Thus, the accumulated liquids may completely kill the well, or liquid slugging or churning may take place and gives more chance for liquid accumulation, eventually, the well may die. To solve such a problem, there are several methods of preventing liquid loadup. The first technique is to maintain the production rate from the gas well above its critical gas flow rate. The critical gas flow rate is the flow rate below which some of the liquid can not be lifted to surface; so, it accumulates in the production string. Many authors studied this critical flow rate or critical velocity1–9. Analysis of research conducted in this area indicated that it is possible to prevent load-up if the gas production rate (velocity) maintained above its critical velocity. This could be achieved by selecting proper tubing size or maintaining low wellhead pressure. Predicting that minimum gas flow rate is very crucial.

Other methods for preventing load-up use continuous liquid removal, as by installing a device in the well such as submersible pump, gas lift injection, plunger lift, tubing inserts, etc. This method requires higher installation and operation costs as compared to using the natural lifting energy of the reservoir. Some other methods of lifting the liquid use chemicals such as surfactants for foam lifting (applied as soap sticks or by annular injection). The surfactant will interact with the water that accumulates at the bottom of a well, creating foam as gas bubbles through it. Lifting the foam is easier than lifting water10.

ANNs are biologically inspired non-algorithmic, nondigital, massively parallel distributive and adaptive information processing systems. They resemble the brain in acquiring knowledge through learning process, and storing knowledge in inter-neuron connection strengths. This paper presents an Artificial Neural Network (ANN) model for predicting the minimum flow rate for continuous removal of liquids from the wellbore. The model is developed using published field. These data were used to train, cross validate and test a three-layer backpropagation neural network model.

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