ANFIS algorithm for mapping computational data of water reservoir homogenization with air bubble flows

Area of Science:

• Chemical Engineering
• Fluid Dynamics
• Artificial Intelligence

Background:

• Bubble column reactors (BCRs) are crucial for liquid homogenization and mixing.
• Traditional computational fluid dynamics (CFD) simulations for BCRs are complex and time-consuming.
• Accurate prediction of air vorticity is essential for optimizing BCR performance.

Purpose of the Study:

• To develop a simplified method for predicting air vorticity in a 3D bubble column reactor.
• To investigate the accuracy and efficiency of an Artificial Intelligence (AI) approach using adaptive networks and fuzzy inference systems (ANFIS).
• To establish correlations for air vorticity prediction that can replace CFD simulations.

Main Methods:

• A 3D bubble column reactor filled with water was simulated using CFD.
• An Artificial Intelligence algorithm, specifically ANFIS with Gaussian membership functions, was employed.
• The AI model was trained using air velocity, pressure, and directional data (x, z) as inputs, with air vorticity as the output. The number of membership functions and input parameters were varied during training.

Main Results:

• AI model accuracy increased with a higher number of membership functions and input parameters.
• The developed AI model achieved high accuracy, with results showing excellent agreement with CFD simulations (regression coefficient near 1).
• The study identified optimal AI parameters: five membership functions and four input variables (air velocity, pressure, x, and z directions).

Conclusions:

• Developed correlations based on AI can accurately predict air vorticity in bubble column reactors.
• This AI-driven approach offers a significantly simpler and faster alternative to traditional CFD simulations.
• This research presents novel correlations for air vorticity prediction in BCRs, addressing a gap in existing literature.