Modelling of liquid loading in gas wells using a software-based approach

Abstract Liquid loading is the most common operational problem influencing gas well productivity for petroleum operator. defined as an constraint that associated with wells where major driving mechanism hydrocarbon production by gas-driven mechanisms. occurs when liquid accumulated in tubing or casing results velocity lower than critical value (the minimum required to push out of well), which overtime leads a hydrostatic back pressure greater formation well, thereby limiting flow into well. The continuous build-up from eventually minimizes and expensive work over operations. However, current mathematical models predict are flawed varying inaccuracies depending on deployed will ultimately lead loss time productivity. In our we present prediction using software-based model incorporating particle swarm optimization algorithm, genetic artificial neural network Bayesian algorithms applications. research findings show simulated accuracy 93% 92% test trained data, respectively, outperformed data-driven sensitivity 83%, algorithm 89% 83%. was postulated robust because its simplicity shown have 77% 73% train respectively. Thus code environment developed presented this paper may resolve many deficiencies gaps technical literature high precision offering saving millions dollars operators.