Experimental and mathematical modelling of biowaste gasification in a bubbling fluidised bed reactor

The production of clean, economically affordable energy with minimum impact on the environment is posing the most significant challenge of the 21st century. According to World Bank data, about 4 billion tonnes of waste is generated per year, out of which cities’ alone contribute 1.3 billion tonnes of solid waste. This volume is forecast to increase to 2.2 billion tonnes by 2025. Three-quarters of this waste is disposed of in landfills while only one-quarter is recycled. In addition to this livestock production is among the most rapidly growing sectors of the agricultural economy, driven primarily by the growing demand for animal protein. Waste from this sector presents a massive disposal problem throughout the world while also offering the potential of being a source of energy and nutrients. In this context, biowaste, originating from households and animal production, could contribute significantly to achieving the ambitious goal set by EU’s Renewable Energy Directive, which commits the EU to produce 20% of its total primary energy demand from the renewables by 2020. This thesis is focused on optimising the bioenergy production from biowaste by employing a thermochemical conversion technology (gasification). To date gasification of municipal solid waste (MSW) and poultry litter has not been explored. In this work, experiments and process simulations are performed which demonstrates the feasibility of biowaste gasification. An additive (limestone) is blended with poultry litter during its gasification in a bubbling fluidised bed to prevent bed agglomeration. Furthermore, a pseudo equilibrium based model (PEM) complemented with experimental data is developed to predict the product gas composition, lower heating value (LHV) and optimised process parameters using Aspen Plus. The carbon conversion, CH4 formation and the equilibrium factor for the water-gas shift reaction were corrected based on experimental data from the gasifier. Data-driven modelling approaches are also developed for simulating the fluidised bed gasification of MSW. Genetic programming (GP) and an artificial neural network (ANN) model using Monte Carlo simulation are exploited. Published experimental datasets are used for model training and validation purposes. A genetic programming based modelling strategy provides analytical expressions which can deliver an insight into determining if the selected feedstock would be an appropriate fuel for gasification or not. A simple and rigorous ANN approach is also developed for deciding hidden layers, number of neurons in the hidden layer and activation function in a network using multiple Monte Carlo simulations. These data-driven models have shown better predictive accuracy over the previously developed PEM. Another advantage of these models is their flexibility to be easily extended for other types of feedstocks/process.