Operational management of fast pyrolysis process using numerical modelling

Depletion of fossil fuels and their environmental impacts such as global warming and climate change are strong motivations for developing alternative energy sources. However, the transition from petroleum-based to renewable infrastructure is challenging. Some of the determining factors to be looked upon of these challenges include but certainly do not restrict only to established fossil fuel infrastructures, availability, viability and economical-feasibility of adequate and appropriate technologies. Bio-based fuels can provide a bridging/alternative to petroleum fuels due to similarity in physical and thermal properties and abundant biomass feedstock. Pyrolysis process; which is an oxygen-free thermal decomposition process is a promising route to convert raw biomass into high-energy-density bio-oil and combustible gases. Pyrolysis can be broadly categorised into three types according to temperature and residence time/heating rates; slow, fast, and flash pyrolysis. The highest bio-oil yield is associated with fast pyrolysis (50‚-75 wt%). Fluidised bed reactor and auger reactor are the most attractive technologies and feasible options for scale-up. Although the experimental test is inevitable for process optimisation and scale-up, computational fluid dynamics (CFD) is a powerful tool for the study of the processes that operate in severe conditions, e.g. high temperature and pressure. In such circumstances, numerical simulation is a much more feasible and affordable approach than experimental tests for parametric study. This study aims to implement a numerical framework to simulate the process that can be used as a design tool; determine the critical operating conditions for process optimisation and scale-up of the pyrolysis reactors; determine the range of conditions for the highest amount of bio-oil yield, and use these results to optimise the lab-scale bubbling fluidised bed reactor and a pilot-scale auger unit. This thesis presents the CFD simulation of a fast pyrolysis process in a two-dimensional standard lab-scale bubbling fluidised bed reactor. In the CFD model, the Eulerian-Granular approach is adopted to address the multi-phase fluid dynamics of the fast pyrolysis processes. A coupled Multi-Fluid Model (MFM) and a chemical solver is used to describe the chemical kinetics. The developed model is validated first using published experimental data then implemented to a parametric study. The impact of different operating factors on the product yields is then presented. The operating conditions are operating temperature, biomass feedstock, biomass feed rate and size, sand particle size, carrier gas velocity and biomass injector location. The results showed that the optimum value of operating temperature is in the range of 500‚-525 ¬∞C. The residence time and secondary reactions can be minimised by increasing the speed of the carrier gas and raising the location of the biomass injector. The intraparticle temperature gradient is lower for smaller biomass particles which resulted in higher heating rates. When larger sand particles are accompanied by higher carrier gas velocity, bio-oil yield increased. Preheating the virgin biomass improves bio-oil yields, whereas other products' yields remain constant. The feedstocks with high cellulose/hemicellulose to lignin ratio are favourable for the production of bio-oil, whereas the highest amount of biochar yield was achieved from high lignin-content feedstocks. A three-dimensional computational fluid dynamic (CFD) model for simulation of a fast pyrolysis process in a pilot-scale auger reactor is also conducted. The adopted methodology used in the CFD simulation of auger reactor is similar to the bubbling fluidised bed reactor. However, the geometry of the auger reactor is more complicated than that of the bubbling fluidised bed reactor and needs to be modelled in a three-dimensional space. Furthermore, the rotating reference frame (RRF) is adopted to simulate the effect of the rotating screw in the auger reactor to avoid the complexity of dynamic mesh generation. The simulation results were compared with the experiment, and the effects of operating conditions were investigated. A parametric study was then conducted to address the effect of operating parameters on the product yields. The results indicate that the optimum temperature for maximum bio-oil production is 500¬∞C. Bio-oil yield rose as the biomass feed flow rate increased due to less vapour residence times, reducing further reaction of the non-condensable fraction in the vapour phase. Nitrogen shows the same impact, enhanced yield due to limited vapour residence time. Increasing the angular velocity of the screw improves the flow of vapours in the reactor. However, this must be balanced against increased unreacted biomass. The simulation gave an optimum of 70rpm for the angular velocity of the screw. The validated CFD model for bubbling fluidised bed reactor is employed to find an optimised set of operating conditions to achieve the highest amount of bio-oil yield. The CFD parametric study was conducted by considering the effect of the key influencing parameters. Machine learning algorithms (MLAs) are then employed to predict the optimised conditions that lead to the maximum bio-oil yield. A developed support vector regression with particle swarm optimisation algorithm (SVR-PSO) was applied to the CFD datasets to predict the optimum values of parameters. The highest bio-oil yield was then computed using the optimum values of the parameters. The CFD simulation is also conducted using the optimum parameters obtained by the SVR-PSO. The obtained CFD results and the predicted values by MLA compared and a good agreement was achieved.