Artificial neural network based modelling approach for municipal solid waste gasification in a fluidized bed reactor

Area of Science:

• Chemical Engineering
• Computational Science

Background:

• Municipal solid waste (MSW) gasification is a key process for waste-to-energy conversion.
• Predicting gasification performance, including lower heating value (LHV) and syngas yield, is crucial for process optimization.
• Fluidized bed reactors are widely used for MSW gasification.

Purpose of the Study:

• To develop and validate artificial neural network (ANN) models for predicting MSW gasification performance.
• To optimize ANN architecture for accurate prediction of lower heating value of gas (LHV), lower heating value of gasification products (LHVp), and syngas yield.
• To assess the viability of ANNs for modeling fluidized bed gasifier performance.

Main Methods:

• Multi-layer feed forward neural networks were employed.
• The Levenberg-Marquardt (LM) back-propagation algorithm was used for training.
• Cross-validation and Monte Carlo runs were utilized for model validation and architecture optimization.
• Nine input and three output parameters were used with experimental datasets.

Main Results:

• ANN models demonstrated high predictive accuracy for LHV, LHVp, and syngas yield.
• Optimal network architectures were identified through rigorous selection procedures.
• The developed ANN methodology proved effective in predicting fluidized bed gasifier performance.

Conclusions:

• ANNs provide a viable and accurate alternative for predicting MSW gasification performance.
• The study confirms the potential of ANNs in optimizing fluidized bed gasifier operations.
• This approach can aid in the efficient management of waste and energy production.