Abstract Nitrogen is an effective agent to be considered in gas injection processes as a part of enhanced oil recovery (EOR) process. Successful design and implementation of nitrogen flooding scenarios greatly depend on accurate prediction of minimum miscibility pressure (MMP). This work highlights the application of three models namely GA-RBF, PSO-ANFIS and MLP for estimation of pure and impure MMP value in nitrogen-crude oil systems based on reliable and valid experimental data covering a wide range of operational parameters gathered from literature. Various statistical and graphical methods were utilized to evaluate the performance of the developed models. The results of the developed models are also compared with reported literature methods for estimation of nitrogen MMP. Results show that the proposed GA-RBF, CHPSO-ANFIS and MLP models are accurate for estimation of experimental data with overall average absolute relative errors of 1.9%, 2.92% and 3.59% and overall R2 values of 0.9847, 0.9686 and 0.9485, respectively. Moreover, the proposed models successfully outperform other literature methods and present more accurate and reliable results.

中文翻译：

---

使用不同计算方案估算 N2 驱提高石油采收率 (EOR) 过程中的最小混相压力 (MMP)

摘要 作为提高石油采收率（EOR）过程的一部分，在注气过程中，氮气是一种需要考虑的有效试剂。氮驱方案的成功设计和实施很大程度上取决于对最小混相压力 (MMP) 的准确预测。这项工作强调了三种模型的应用，即 GA-RBF、PSO-ANFIS 和 MLP，基于可靠和有效的实验数据，涵盖从文献中收集的各种操作参数，用于估计氮-原油系统中纯和不纯的 MMP 值。使用各种统计和图形方法来评估开发模型的性能。还将开发模型的结果与用于估计氮 MMP 的文献方法进行比较。结果表明，提出的 GA-RBF，CHPSO-ANFIS 和 MLP 模型对于估计实验数据是准确的，总体平均绝对相对误差分别为 1.9%、2.92% 和 3.59%，总体 R2 值分别为 0.9847、0.9686 和 0.9485。此外，所提出的模型成功地优于其他文献方法，并提供了更准确和可靠的结果。