Advanced combustion kinetics models are of widespread use to predict carbon losses from coal combustion. However, those models cannot completely capture the complexity of the real phenomena affecting the fluid flow in a full-scale utility boiler, such as burner-to-burner interactions and bottom hopper vortexes or reversed-flows, and usually underpredict carbon in ash values. The use of CFD codes offers a more detailed treatment of the fluid dynamics involved in the boiler. However, most of them do not incorporate advanced kinetics submodels for char oxidation. In this paper, rank-dependent correlations and ash inhibition submodel have been coupled to a commercial CFD code, significantly improving carbon in ash predictions. Results from the simulation of the ASM Brescia power plant (Italy) for three different South-American coals are compared against plant laboratory values, using either the popular single film combustion model or the modified combustion model discussed in this paper. (c) 2005 Elsevier Ltd. All rights reserved.