Optimized Artificial Neural Network for Evaluation: C4 Alkylation Process Catalyzed by Concentrated Sulfuric Acid

Area of Science:

• Chemical Engineering
• Computational Chemistry
• Process Optimization

Background:

• The C4 alkylation process is crucial for producing high-octane gasoline components.
• Optimizing product distribution and octane number is essential for economic viability and environmental compliance.
• Traditional modeling approaches may struggle with the complexity of reactor dynamics and feedstock variations.

Purpose of the Study:

• To develop and optimize an artificial neural network (ANN) model for evaluating product distribution and octane number in sulfuric acid-catalyzed C4 alkylation.
• To investigate the influence of feedstock composition, operating conditions, and reactor type (stirred tank vs. rotating packed bed) on the process outcomes.
• To identify the optimal ANN architecture and parameters for accurate prediction.

Main Methods:

• An artificial neural network model was designed and optimized.
• Input parameters included feedstock compositions, operating conditions, and reactor types.
• Network topology (10-20-30-5), Bayesian Regularization backpropagation, and tan-sigmoid transfer function were selected as optimal.
• Research octane number and product distribution were the output parameters.

Main Results:

• The optimized ANN model achieved a training mean square error of 5.8 × 10^-4 and a testing mean square error of 8.66 × 10^-3.
• The model demonstrated high accuracy with a correlation coefficient of 0.9997 and a ±22% deviation.
• Parameter analysis revealed the impact of operating conditions on octane number and product distribution.

Conclusions:

• The developed ANN model effectively predicts product distribution and octane number for C4 alkylation in both stirred tank and rotating packed bed reactors.
• The study highlights the potential of ANNs for evaluating and optimizing complex chemical systems.
• This approach offers a pathway for enhanced process control and efficiency in alkylation processes.