Investigation of "Whole Tree" combustion in a packed-bed

Abstract

A shrinking core model of the combustion of individual chunkwood and particle wood elements is developed and validated by comparison with literature data. The model is formulated on the physical evidence that large wood specimens inserted into a hot environment lose weight mostly over a relatively thin outside layer, while the interior (core) remains relatively undisturbed. The modeling of the complete process requires a correlation of the turbulent heat and mass transfer coefficients which includes the effects of transpiration of volatilized organic compounds and moisture, geometry, and shrinking radius. The model shows that under boundary layer diffusion control, the extemal boundary layer thickness and diffusional characteristics are constantly modified by the effects of blowing. Hence, for green wood specimens, the cooling effects of transpiration and the latent heat of evaporation do slow the buming rate contrary to earlier publications. The mass and heat transfer coefficients are then modified and incorporated into the analysis of a steady state one dimensional, packed-bed model of "whole tree" combustion. The bed is assumed to be loaded uniformly from the top, countercurrent to the preheated air stream of superficial velocity Vs' The preheat time of the fuel elements is assumed negligible. The mass loss rate is formulated in a quasi-steady shrinking core submodel with a uniform core temperature approximation and negligible heat of pyrolysis. The model is simple, and yet reliable enough to predict adequately the bumout time and the depth of the combustion zone of a 100 MW "whole tree" facility, by showing that fuel elements properties (size, moisture content), strongly influence combustion characteristics, especially that higher moisture produces depressed flame temperature, longer bumout time, and larger values· of combustion zone. The uniform core temperature approximation and negligible heat of pyrolysis assumption of the quasi-steady model are latter replaced by a transient moving pyrolysis front submodel. A modified Variable Grid Method called Continuous Mapping Variable Grid Method (CMVGM) is developed to solve the two-phase moving boundary problem in which in addition to the 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 moving interface, the external boundary of the solid fuel is receding due to combustion. With variable time step and variable radiaI grid, the scheme continuously traclcs the position of both moving boundaries together with the position of the entire "whole tree" elements inside the bed, by continuously mapping in the core and shell regions, the computationaI domain of time step (n+1) to that of the previous time step (n). The results of the transient model show that indeed, the preheat time is negligible. As anticipated, the core temperature also remains approximately constant during most of the combustion process, but close to the end, it rises steadily toward the interface temperature. This is speculated to be due to the effect of "cooking" believed to be caused by rising internaI pressure and fuel temperature carried inward by convective moisture and pyrolysis gases. This phenomenon, however, does not aIter significantly the results of the quasi-steady model which tends to give slightly lower burnout time (3 to 10 %) and higher depth of combustion. zone ( 10 to 25 %)