Artificial Neural Network predictive model for Hydrogen production using Biomass gasification in a pilot plant

The growth in population nowadays has led to an increase in the consumption of the fossil fuels like oil and gas, which leads to depletion and shortage in the supply of the oil and gas. Also it will lead to an increase in the pollution and greenhouse effects in the environment. The need for a reliable, affordable and clean energy supply rises as it is very important for society, economy and the environment. Hydrogen production from biomass gasification is considered a very promising clean energy option for reduction of greenhouse gas emissions and energy dependency. The complexity of the biomass gasification process has led the researchers to develop models to simplify the process and save time and energy. A lot of models have been developed like the equilibrium model, kinetic model and the Artificial Neural Network (ANN) model. ANN models are simple to use, easy to generate and require a short period of time to get acceptable results depending on the pool of previous experimental data comparing to the other models that need power, time, a lot of assumptions and calculations to obtain good results. The main objectives of this study are: 1- to design and develop an Artificial Neural Network (ANN) model for the hydrogen production from biomass gasification process. 2- To evaluate the results of the model and validate them with the previous experimental data. 3- To compare the results of the simulation with different ANN models with the SIMCA-P software model. To achieve the goal of this study, four (4) ANNs have been developed after performing a preliminary analysis which was done by SIMCA-P11 and SIMCA-P13 software to determine the factors that affect the hydrogen production and also as it has a linear modelling for the process which is compared to the results of the ANNs. ANNs performed better in the prediction process with a mean squared error (MSE) of 5.4%. This validate that the ANN modelling is better for the purposes of prediction comparing to the other models available