This paper presents an investigation of wood combustion in a laboratory-scale fixed bed with an aim of establishing the effect of CO2 environment on flame propagation speed and flame structures. Different oxy-fuel combustion atmospheres in which the composition of O-2 in CO2 was varied from 21% to 50% by volume were tested and compared to air-fuel condition. Euler-Lagrange (Computational Fluid Dynamics - Discrete Element Method, CFD-DEM) approach which captures information of individual particle processes is used to model wood conversion in a packed bed. Results show that flame front propagation speed in oxy-fuel atmosphere reduced to 78% of that of the air-fuel condition with similar O-2 concentration. For oxy-fuel conditions, propagation speed increased with increase in O-2 concentration. The CFD-DEM model agrees very well with experimental values for mass loss, propagation speed and flame front positions. However, peak temperatures are poorly predicted at lower oxygen concentrations. The accuracy of temperature prediction improves at higher oxygen concentrations. During initial and devolatilization stage, mass fraction of tar predicted in CO2 environment are smaller than in N-2 environment, while the amount of CO predicted is almost equal in both environments. However, during char combustion stage a high amount of CO is observed in oxy-fuel conditions.