OPTIMISATION OF BIOMASS GRATE FURNACES WITH A NEW 3D PACKED BED COMBUSTION MODEL - ON EXAMPLE OF A SMALL-SCALE UNDERFEED STOKER FURNACE Mehrabian, R. 1,2,* , Scharler, R. 1,2,3 , Weissinger, A. 4 , Obernberger, I. 1,2,3 1 BIOENERGY 2020+ GmbH, Inffeldgasse 21b, 8010 Graz, Austria Tel.: +43 (0)316 8739232; Fax: +43 (0)316 8739202; E-mail: ramin.mehrabian@bioenergy2020.eu 2 Institute for Process and Particle Engineering, Graz University of Technology, Inffeldgasse 21b, A-8010 Graz, Austria 3 BIOS BIOENERGIESYSTEME GmbH, Inffeldgasse 21b, A-8010 Graz, Austria 4 KWB - Kraft und Wärme aus Biomasse GmbH, Industriestrasse 235, A-8321 St. Margarethen, Austria ABSTRACT: The design and optimisation of a biomass grate furnace requires accurate and efficient models for the combustion process on the grate as well as the turbulent reactive flow in the combustion chamber. Computational Fluid Dynamics (CFD) have been successfully applied for gas phase combustion. However, no numerical models for the biomass packed bed combustion, which can be used as engineering design tools, are commercially available at present. This paper presents an innovative 3D CFD model for biomass packed bed combustion consisting of an Euler- Granular model for hydrodynamics of gas-particle multiphase flow and a thermally thin particle model for combustion of biomass particles. Modelling the particle trajectories and the thermal conversion of each particle in the bed constitutes the simulation of the entire bed combustion. The simulation of a small-scale underfeed stoker furnace of KWB has been successfully performed by the application of the new packed bed combustion model. The positions of the drying, pyrolysis and char burnout zones in the fuel bed as well as the temperature distribution among the particles seem to be plausible and could be confirmed by observations. Furthermore, a good qualitative agreement concerning the flue gas temperatures measured by thermocouples at different positions in the combustion chamber, and CO emissions measured at boiler outlet could be achieved. The new packed bed model provides the advantages of considering the release profiles of species and energy from the fuel bed close to reality and enables to consider the chemical compositions, size and physical properties of the fuel particles as well as the influence of primary air distribution and grate motion on the particle trajectories. Keywords: biomass, combustion, fixed bed, CFD, modelling. 1 INTRUDUCTION AND OBJECTIVES CFD simulation techniques are an efficient tool for the design and optimisation of biomass grate furnaces. They have demonstrated to be valuable to predict the flow and the gas phase combustion in furnaces [1-4]. However, at present there is a lack of reasonably accurate and computationally efficient simulation tools for packed bed biomass combustion. The main problems encountered in modelling biomass packed bed combustion are the hydrodynamics of the gas-solid multiphase flow and thermal conversion of the biomass particles. There are various simulation methods applicable to the dense gas-solid multiphase flows (granular flows). Generally they can be classified into two approaches: the discrete element methods (DEM) based on the molecular dynamics and the continuum mechanics methods or two-fluid model (TFM) based on the assumption that the gas and particulate phases form two inter-penetrating continua [5]. The discrete element method is fully based on the Lagrangian framework, i.e. the motion of each particle is defined by classical Newtonian mechanics and contact mechanics of deformation. The particle-particle collision is modelled by the soft sphere method [6] or hard sphere method [7]. In general, discrete models are powerful and they are able to predict the rotation and velocity of each particle. Moreover, they allow the investigation of the effect of individual physical particle properties in the granular flow. The limiting parameter in the DEM is the number of particles. Hence, in most cases, the calculation time is too high for industrial scale systems. The Euler/Euler two-fluid model assumes that the particulate phase behaves as a fluid. Therefore, the continuity and momentum equations with jump conditions for phase interfaces are solved for both gas and particulate phases. In this approach all the particles are assumed to be identical, specified by their mean diameter and density. Therefore handling a poly-disperse system, i.e. a system with different particle sizes, requires several solid phases corresponding to the number of particle diameter classes. Hjertager reported a quadratical increase of computational effort with the number of phases [8]. Additionally, the modelling of particle- particle collisions in this approach is rather complicated. It has been implemented in the momentum equation of solid phase by the viscosity and normal stress tensor of the solid phase. There are several correlations for the solid viscosity term and the solid normal stress tensor [9- 14]. They have been driven by making an analogy between the particle-particle collisions and the kinetic theory of gases [15]. The concept of granular temperature is defined to represent the kinetic energy of random particle fluctuations around their mean velocities. A conservation energy equation is formulated for this kinetic fluctuation energy in which the kinetic energy is produced by shear and fluid turbulence and dissipated by inelastic collisions and interaction with the fluid. The collisions between the particles are assumed to be a function of this kinetic fluctuation energy. The capability of the TFM for simulation of granular systems has been proven by its numerous applications, see [16] and its references. In the present work the commercial CFD software, ANSYS FLUENT 12, has been utilised to simulate the hydrodynamics of the packed bed granular flow. Among the available gas-solid multiphase models in ANSYS FLUENT the Euler-Granular model has been selected because it is based on the kinetic theory of granular flows and allows the consideration of inter-particle interactions which are of key importance in modelling of packed beds. This model has been successfully used for predicting the hydrodynamic behaviour of a bubbling fluidised bed [17-19]. In this study the Euler-Granular 18th European Biomass Conference and Exhibition, 3-7 May 2010, Lyon, France 1175