Using a Proxy Model Network to Predict Liquid Loading

The gas well loading phenomenon is one of the most serious problems that reduces, and eventually cuts, production in gas wells. This phenomenon occurs as a result of liquid accumulation of condensate inthe well bore. Over time, these liquids cause additional hydrostatic backpressure on the reservoir which results in a continual reduction of the available transport energy. The well therefore starts slugging whichgives an even larger chance of liquid accumulation that completelyovercomes the reservoir pressure and causes the well to die [1]. Typical solutions of loading phenomenon were to unload the well artificially; either mechanically using pumps or with gas lift kicking with nitrogen through coiled tubing. However, in addition to the expense and loss of production, artificial lift solutions remain temporary and the well is subject to reloading again. Therefore, thought was directed toward developing some solutions that enable the well to continuously unload itself without the aid of external help unloading operations [2].