A novel PSO-LSSVM model for predicting liquid rate of two phase flow through wellhead chokes

• An artificial intelligent model is developed for predicting liquid flow rate of wellhead choke from production data.
• Least square support vector machine (LSSVM) is implemented for this purpose.
• Particle swarm optimization (PSO) is applied for optimizing LSSVM tuning parameters.
• Proposed model has more accuracy than other former correlations.

Two-phase flow through chokes is common in oil industry. Wellhead chokes regulate and stabilize flow rate to prevent reservoir pressure declining, water coning and protecting downstream facilities against production flocculation. Choke liquid rate prediction is a basic requirement in production scheme and choke design. In this study, for the first time a least square support vector machine (LSSVM) model is developed for predicting liquid flow rate in two-phase flow through wellhead chokes. Particle swarm optimization (PSO) is applied to optimize tuning parameters of LSSVM model. Model inputs include choke upstream pressure, gas liquid ratio (GLR) and choke size which are surface measurable variables. Calculated flow rates from PSO-LSSVM model are excellently consistent with actual measured rates. Moreover, comparison between this model and related empirical correlations show accuracy and superiority of the model. Results of this work indicate PSO-LSSVM model is a powerful technique for predicting liquid rate of chokes in oil industry.