Loading...

Modeling relative permeability of gas condensate reservoirs: Advanced computational frameworks

Mahdaviara, M ; Sharif University of Technology | 2020

797 Viewed
  1. Type of Document: Article
  2. DOI: 10.1016/j.petrol.2020.106929
  3. Publisher: Elsevier B. V , 2020
  4. Abstract:
  5. In the last years, an appreciable effort has been directed toward developing empirical models to link the relative permeability of gas condensate reservoirs to the interfacial tension and velocity as well as saturation. However, these models suffer from non-universality and uncertainties in setting the tuning parameters. In order to alleviate the aforesaid infirmities in this study, comprehensive modeling was carried out by employing numerous smart computer-aided algorithms including Support Vector Regression (SVR), Least Square Support Vector Machine (LSSVM), Extreme Learning Machine (ELM), Multilayer Perceptron (MLP), Group Method of Data Handling (GMDH), and Gene Expression Programming (GEP) as predictors and Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Levenberg-Marquardt Algorithm (LMA), Bayesian Regularization (BR), Scaled Conjugate Gradient (SCG) and Randomized Polynomial Time (RP) as optimizers. To this end, a wide variety of reliable databanks encompasses more than 1000 data points from eights sets of experimental data was utilized in the training and testing steps of the modeling process. The predictors were integrated with optimization algorithms to assign the optimum tuning parameters of each model. The modeling was implemented in two different manners from the standpoint of models inputs: (1) 2-input (saturation and capillary number); (2) 3-input (saturation, interfacial tension, and capillary number). The results of the comparison between these strategies demonstrate more accuracy of the models when employing three independent parameters as the input (3-input). Among the developed models, the MLP-LMA modeling algorithm outperformed all other models with root mean square errors (RMSEs) of 0.035 and 0.019 for gas and condensate phases, respectively. At the end, in a comparison between both 2-input and 3-input MLP-LMA models and five traditional literature models, both smart modeling approaches were established themselves as the most accurate techniques for estimation of relative permeability in gas condensate reservoirs. © 2020 Elsevier B.V
  6. Keywords:
  7. Comprehensive modeling ; Empirical correlation ; Artificial intelligence ; Capillarity ; Computer aided instruction ; Computer programming ; Conjugate gradient method ; Data handling ; Gas condensates ; Gas permeability ; Gases ; Gene expression ; Genetic algorithms ; Least squares approximations ; Mean square error ; Particle swarm optimization (PSO) ; Petroleum reservoir engineering ; Petroleum reservoirs ; Polynomial approximation ; Polynomial regression ; Support vector machines ; Comprehensive model ; Empirical correlations ; Gene expression programming ; Group method of data handling ; Least square support vector machines ; Levenberg-Marquardt algorithm ; Support vector regression (SVR) ; Learning systems ; Correlation ; Empirical analysis ; Gas condensate ; Hydrocarbon reservoir ; Permeability
  8. Source: Journal of Petroleum Science and Engineering ; Volume 189 , June , 2020
  9. URL: https://www.sciencedirect.com/science/article/abs/pii/S0920410520300280