In this study, pre-processor code based on structural behavior of coal is applied to predict yields, pyrolysis rate and compositions of volatile and char. These parameters are used in the devolatilization and char burnout submodels as user-defined functions of commercial CFD code. The predicted characteristics of these sub-models are compared with those employing the conventional model based on experiment and validated against the measurement of a 2.1 MW swirling pulverized coal flame in a semi-industrial scale furnace. And the influence of the turbulence-chemistry interaction on pulverized coal combustion is analyzed.