Abstract
M.Ing. (Mechanical Engineering)
This work aims to develop a Computational Fluid Dynamics (CFD) model of a domestic biomass briquette combustor as a means of providing a platform to enhance combustion performance in such combustion systems. In addition to the numerical modelling which forms the core of the project, experimental tests were conducted as a means of evaluating model accuracy. The combustor used in the experiment has a cylindrical combustion chamber of length 25cm and a diameter of 11cm.
A variety of fuels were used in the experimental tests conducted. However, the fuel chosen for use for purposes of model validation is a peanut shell-cow dung mixture briquette formed under a 5 ton load. The proximate analysis reveals the constituent moisture, medium volatile and combustible matter and ash contents to be 3.44%, 64.99% and 31.57% respectively. On the other hand, the calorific value is determined to be 17.1 MJ/kg.
Due to resource constraints, only devolatilisation is modelled. This decision is precipitated by the determination made by the proximate analysis showing the volatiles to be the most abundant constituent of the fuel. The computational domain of the model represents the combustion chamber which is separated into two regions-namely, freeboard and fuel bed. The freeboard consists only of the gas phase while the fuel-bed consist of multiple phases in interaction, that is, the solid fuel and gas phases. The multiphase nature of the model is captured by an Eulerian-Eulerian model where the fuel region is considered a packed bed with the solid fuel represented by a granular phase.
The combustion modelled is based on the homogeneous reactions taking place in the freeboard which transfer heat to the solid fuel to initiate and sustain mass transfer of reactants to the freeboard. The exchange of heat between the gas and solid phases is carried out through radiation and convection. However, the inability of the commercial CFD package in use (Ansys Fluent 17.0) to calculate multiphase heat transfer necessitates the introduction of User Defined Functions (UDFs) to capture the radiation and convection contributions to the source term of the solid fuel energy equation.
The rate of volatile species mass transfer, on the other hand, is determined by an iterative process. The constant species volatile fractions for the volatile fuel component of a woody biomass obtained from literature are used as the upper limit mass transfer rates for the respective species. The lower limit is based on the estimated average mass loss rate determined from the experimental mass decay curve. The proportions of the species were kept the same for each iteration. The values of mass transfer rates of each iteration were determined by dividing the species mass transfer rates of the upper limit with the nth integer iteration number. Comparisons carried out at 60 seconds of combustion between each iteration and the experimental temperatures at the exit as well as within the freeboard (4cm from the exit) along the central axis showed improved predictions with each iteration. As a consequence of resource constraints, the most important being time, only ten iterations could be carried out with each case run for 5 minutes.
The temperatures measured at the exit as well as the freeboard (4cm from the exit) show the freeboard temperatures to be greater than the exit temperatures. The reverse situation is the case for the model predicted temperatures in this region. This relational agreement between the model and experiment is observed only within 3cm of the exit. Also, as modelled, the mass transfer rates specified are consistently higher than those predicted. This shows the specified rates to be upper limits. The predicted rates are also observed to be dependent on the specified values and independent of the species being transferred. Furthermore, the reaction rates for each of the homogeneous reactions are revealed to be heavily dictated by the large-eddy mixing time scale in the longitudinal...