Objective: The accurate combustion simulation of wood is essential for improving fire safety in architectural and wildland contexts. Existing studies, which predominantly rely on fire dynamics simulators (FDS), face considerable limitations, particularly in terms of grid adaptability for curved geometries and the oversimplification of pyrolysis models. These limitations often result in substantial deviations from experimental data, thereby reducing the reliability of fire safety predictions. This study develops a comprehensive simulation framework for wood ignition using ANSYS Fluent to address the above gaps. This framework is validated through controlled experiments to improve its predictive accuracy for fire dynamics. Methods: The experimental phase of this study employed small cylindrical Finnish pine wood blocks, each with a diameter and length of 30 mm. The wood blocks had an average moisture content of 8.68% and an apparent density of 460.27 kg/m³. Thermogravimetric analysis (TGA) was conducted to quantify wood moisture content, which was found to be 8.87%, and pyrolysis conversion rate, which reached 0.745 at 500 ℃. Ignition tests were performed under a heptane flame, revealing mass loss ratios of 20%-50% within just 2 min. This remarkable mass loss was attributed to surface charring and the development of internal pyrolysis gradients. Combustion was further characterized by three distinct stages: an evaporation stage (Stage Ⅰ) marked by slow mass loss; a rapid pyrolysis stage (Stage Ⅱ) defined by accelerated degradation; and a slow mass decline stage (Stage Ⅲ), wherein the accumulation of a char layer inhibited further reactions. Postcombustion analysis highlighted the formation of a uniform 5 mm char layer, with internal conversion rate gradients showing a surface value of 19.14% and low internal values. These gradients were influenced by gas permeability and temperature distribution within the wood. In the numerical simulation phase, ANSYS Fluent was employed to model the complex multiphase processes involved in wood ignition. User-defined functions (UDFs) were developed to incorporate drying and pyrolysis. Wood components were simplified into moisture and organic matter, with porosity values of 0.676-0.679 derived from cell wall density measurements. Pyrolysis kinetics were modeled using a modified Arrhenius model, integrating parameters obtained from TGA. A virtual heat-exchange layer was introduced to adjust surface heating rates, effectively mimicking the insulating effect of water vapor observed in experiments. A rotating slip-grid method ensured the uniform heating of the wood sample. Meanwhile, the large eddy simulation was employed to capture the turbulent combustion of heptane. Radiation effects were modeled using the discrete ordinates approach, which was coupled with energy equations to account for stage changes and chemical reactions. Results: The key innovations of this study include the development of a spatially resolved conversion rate gradient model for char layers with the thickness x, expressed as α_s(x)=e^-0.28-x, and dynamic porosity adjustments to reflect gas transport limitations within the wood. Simulation results demonstrate strong agreement with experimental mass losses, thereby validating the proposed method. This study reveals that surface charring substantially decelerates pyrolysis by reducing gas permeability, whereas internal temperature gradients govern the cessation of reactions within the wood. Conclusions: This work establishes a robust Fluent-based framework for simulating wood ignition, effectively overcoming the limitations of FDS through advanced mesh resolution and detailed pyrolysis modeling. By integrating experimental data into UDFs, the method established herein enhances predictive capabilities for fire spread in structural and environmental fire scenarios. Future research could focus on expanding the model to incorporate heterogeneous secondary reactions, thereby further bridging the gap between simulations and real-world fire behavior.