Modeling of steam distillation mechanism during steam injection process using artificial intelligence

Area of Science:

• Petroleum Engineering
• Chemical Engineering
• Artificial Intelligence

Background:

• Steam distillation is a crucial mechanism in thermal oil recovery.
• Accurate simulation of steam distillation is vital for optimizing oil extraction.
• Existing simulation methods may require complex inputs or exhibit limitations.

Purpose of the Study:

• To simulate steam distillation processes using artificial intelligence (AI) tools.
• To evaluate the predictive performance of artificial neural networks (ANN) and adaptive neurofuzzy inference system (ANFIS) for distillate recovery.
• To compare AI-based simulations with traditional equation of state methods.

Main Methods:

• Utilized sixteen crude oil datasets from existing literature.
• Trained ANN and ANFIS models using thirteen datasets.
• Validated model performance on three independent test datasets.
• Employed modified Peng-Robinson equation of state for comparative simulation.

Main Results:

• Developed AI models accurately predict distillate yield based on oil properties.
• ANFIS model exhibited lower prediction errors on both training and test datasets compared to ANN and equation of state methods.
• The models demonstrate high compatibility with input oil properties for efficient prediction.

Conclusions:

• ANFIS provides a highly accurate and efficient method for simulating steam distillation in oil recovery.
• AI tools offer a promising alternative to traditional methods for predicting distillate yield.
• The developed models can significantly aid in optimizing thermal oil recovery processes.