Many of the pyrolysis kinetics for biomasses available in the literature are based on idealised particle shapes (sphere, cylinder, cuboid/cube) or very small particle sizes to describe only the kinetics of the chemical reactions. However, actual fuels, such as those used in fluidised bed gasification plants, consist of significantly larger particles, which lead to transport limitations. Consequently, detailed particle models that account for heat conduction, particle porosity, the degree of char formation from pyrolysis of raw biomass, etc., are needed. However, due to their complexity, models describing such apparent kinetic mechanisms are hard to implement in process simulations, especially if fuel particles have a large particle size distribution.
Hence, simple empirical kinetic models for commercially available wood chips were developed in the present work. Different particle sizes were distinguished from a sieve analysis based on the size fractions 1.18 mm, 3.15 mm, 7.1mm and 10 mm. Since the sieve residue of the 1.18 mm size fraction consisted mainly of spruce needles, only those were used for the pyrolysis tests. The influence of the pyrolysis temperature on the individual pyrolysis products, namely gases, tars, water, and residual char, is considered in detail in the temperature range between 600 °C and 800 °C. In the case of the gases, a further distinction was made based on the gas components H2, CO, CO2, CH4 and other non-condensable hydrocarbons. The composition of the tars was investigated in detail for selected test points by analysing 25 individual components.
The fluidised bed test facility requires a sufficiently large reactor volume for the experiments with large particle fractions. This posed a challenge for the derivation of pyrolysis kinetics, as it results in a non-negligible influence of the residence time distribution on the released pyrolysis products. Thus, the gas release rates measured at the reactor outlet represent a superposition of the actual gas release at the particle and the plant behaviour. An algorithm was developed that allows subtracting the system behaviour of the plant from the measured values to obtain the gas release rate directly at the particle’s surface. For this purpose, the numerical solution of the transient convection-diffusion equation is used in a global optimisation algorithm, and the release rate of the pyrolysis products at the particle is approximated via a polygon course of 32 freely selectable support points. Thus, high gradients and abrupt changes of the release rate at the beginning of the pyrolysis can be represented well, and, unlike in inverse calculations, no smoothing is necessary.
Simple kinetic models were derived for the four particle size fractions (1.18mm to 10mm) using the determined gas release rates. These kinetic models describe both the heating process and the actual pyrolysis using a maximum of four model parameters, and therefore, they can be easily implemented in process simulations.