Application of hybrid neural particle swarm optimization algorithm for prediction of MMP

Carbon dioxide (CO2) injection is one of the most effective methods to improve enhance oil recovery. While the local displacement in CO2 injection process is highly dependent on minimum miscibility pressure (MMP), so this is one of the main factors in design of CO2 injection operations. There are several experimental methods utilized to determine MMP such as slim tube displacement and rising bubble apparatus (RBA); however, these methods are expensive and time consuming. On the other hand, computational methods are being used in the recent decades in order to create inexpensive, rapid and robustness models to estimate gas–oil MMP. In this research, we proposed new artificial neural network (ANN) optimized by particle swarm optimization (PSO) to estimate pure and impure MMP of oils. PSO used to find best initial weights and biases of neural network. As input parameters, neural network considered the reservoir temperature, fluid composition and injected gas composition and MMP as target parameter. The performance of hybrid neural particle swarm optimization model (ANN–PSO) is compared with calculated results for common gas–oil MMP. The results show that proposed model yielded accurate gas–oil MMP with lowest average absolute deviation (AAD) and highest square of correlation coefficient (R2).