Evaluation of Models to Predict Liquid Loading in Gas Wells
MetadataVis full innførsel
Liquid production in the form of water or condensate is essentially the most common problem in mature gas wells. For efficient operation of the well, the produced liquid must be continuously transported to the surface. This is only possible when the gas rate is above a certain threshold known as a critical velocity. With depletion, the gas rate declines, when it reaches below the critical velocity, liquid starts to accumulate at the bottom of the well creating what is referred to as liquid loading. Liquid loading results to rapid gas rate decline, reduction in the gas well ultimate recovery and may completely kill the well if immediate liquid unloading actions are not taken. Several correlations have been devised to predict the inception of liquid loading. Some correlations are based on the liquid droplet reversal and others are based on the liquid film reversal. However, each correlation provides divergent views on the critical gas velocity required to initiate liquid loading. Some correlations under- predicts while others over-predicts the critical gas velocity.In the present work, the liquid droplet model of Turner et al. (1969) and the liquid film reversal models of Barnea (1986,1987), Luo et al. (2014) and Shekhar et al. (2017) were compared against the field data shared in the literatures. It was observed that the Turner s model can accurately predict the onset of liquid loading and the loading status of gas wells producing under stable conditions while the model was found to under-estimate the critical velocity for wells producing under unstable condition. It was also observed that, out of all film reversal models compared, the Shekhar et al. (2017) model can better predict the critical velocity especially for deviated wells with large diameters in the range of 2 to 6 inches.The study on the effect of droplet entrainment in the film reversal models have revealed that the film reversal models can over-estimate the critical velocity by up to 18% if the fraction of droplets entrainment is neglected in the models. The Shekhar et al. (2017) was therefore modified by incorporating the (Oliemans et al. 1986) entrainment correlation instead of the Wallis (1969) correlation used in the Luo et al. (2014) model. The resulting critical velocity predictions were found to be better than the that of the original Shekhar et al. (2017) while that of Luo s model were still conservative. The study on the liquid loading criteria along the wellbore have revealed that both the wellhead and the bottomhole should be used as the evaluation point of the critical gas velocity so that a precise conclusion of whether the well is loaded or unloaded could be drawn.