10.4122/1.1000000957
Bengtsson, Peter
Peter
Bengtsson
peter.bengtsson@vxu.se
Claesson, Johan
Johan
Claesson
johan.claesson@chalmers.se
Bengtsson, Peter
Peter
Bengtsson
peter.bengtsson@vxu.se
MEASUREMENT AND MODELING OF PELLET DRYING
8th Symposium on Building Physics in the Nordic Countries
2008
2008
For the production of pellets, one of today’s and tomorrow’s alternative for environmental-friendly heating of small houses, drying of the raw wood material is required. In this work, the drying of sawdust and wood chips is studied both experimentally and theoretically. An experimental bed dryer that dries up to 0.25 m3 wood in batches was constructed and used in the experiments. The purpose is to develop energy-saving drying processes by studying drying parameters such as air temperature, air velocity, bed height and type of wood with different initial moisture contents. The drying rate and final moisture content in the product are key factors. Experiments show that the drying equipment and applied method are applicable to evaluate different drying parameters and their influence on the drying course. The drying is characterized by a distinct drying zone that moves through the bed.
In order to simulate the drying process, a model utilizing coupled moisture and heat balance equations is proposed and solved by using the finite-difference method. There are balances for absorbed water in the pellets, water vapor, air and heat. The model requires an intrinsic mass transfer coefficient for the evaporation, and the thermal conductivity for the air and wood mixture. These key parameters are difficult to determine. However, the coefficients are fitted and estimated in the experimental analysis. As in the experiments, the model shows distinct drying-zone boundaries characterized by large temperature gradients. The large gradients require very fine division in cells in the numerical calculations in order to represent properly the coupled processes. A typical calculation of a drying process during a few hours with 100 cells require around 30 s computer time on an ordinary PC. The model is utilized to predict the drying course for varying drying conditions and material, and it shows good agreement with the experimental data.