Prediction of pressure in different two-phase flow conditions: Machine learning applications

The accurate prediction of pressure has extensive applications in the petroleum industry, especially in the optimization of continuous field production, quantifying reservoir performance, adjusting the cost of oil production and assessment of workover operations. On the other hand, the complexity associated with the twophase flow system makes the computational determination of bottom-hole pressure difficult and timeconsuming, so in the current work, two novel machine learning approaches based on Gradient tree boosting (GTB) and Extreme Learning Machine (ELM) have been proposed. The comparison of 458 actual pressure values and proposed GTB and ELM outputs has clarified that machine learning approaches have excellent performance in oil field calculations with R-squared values of 1 and 0.999 and also mean relative errors lower than 4% for these models. Additionally, sensitivity analysis on the input variables shows that the most effective parameter in the determination of bottom-hole pressure is well-head pressure.

Automated summary

Accurate prediction of pressure is crucial in the petroleum industry for various applications, and machine learning approaches such as Gradient tree boosting and Extreme Learning Machine have been shown to have excellent performance in oil field calculations. Well-head pressure is identified as the most effective parameter in determining bottom-hole pressure.