Hybrid intelligence methods for modeling the diffusivity of light hydrocarbons in bitumen

Area of Science:

• Petroleum Engineering
• Chemical Engineering
• Artificial Intelligence

Background:

• Solvent diffusivity is critical for solvent-assisted processes in the bitumen industry.
• Accurate prediction of light hydrocarbon diffusivity in bitumen is essential for process design and optimization.

Purpose of the Study:

• To develop and evaluate advanced computational models for predicting solvent diffusivity in bitumen systems.
• To enhance the predictive accuracy of the Adaptive Neuro-Fuzzy Inference System (ANFIS) using optimization algorithms.

Main Methods:

• Implemented a novel Adaptive Neuro-Fuzzy Inference System (ANFIS) model.
• Utilized Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) to optimize ANFIS performance.
• Trained and tested models using a comprehensive dataset from existing literature, considering temperature, pressure, and alkane molecular weight as inputs.

Main Results:

• The ANFIS-PSO model demonstrated superior performance with high correlation coefficients (R² = 0.991 for training, R² = 0.987 for testing).
• Statistical analysis confirmed the ANFIS-PSO model's higher desirability compared to ANFIS and ANFIS-GA.
• All models were validated using statistical parameters and graphical methods.

Conclusions:

• The ANFIS-PSO model offers a highly accurate and reliable approach for predicting light hydrocarbon diffusivity in bitumen.
• Optimized ANFIS models significantly improve the evaluation of solvent diffusivity, crucial for bitumen processing.
• This research provides a valuable tool for optimizing solvent-assisted bitumen extraction and processing technologies.