Abstract: Subchannel analysis assumes uniform flow and heat distribution within each subchannel. However, components like guide tubes and mixing grids create heterogeneity, often involve direct application of correction factors, with limited research dedicated to methodologies based on local parameter modifications. The refined subchannel model addresses the cold-wall effect of guide tubes by subdividing the corresponding subchannels, yet the cold-wall effects from the channel box and the peripheral core flow regions remain inadequately considered. To further address the channel box cold-wall effect, a composite refined subchannel model was proposed in this study, building upon the refined model by additionally partitioning the fluid domains near the channel box or core periphery into dedicated subchannels. A turbulent mixing model was developed based on the eddy diffusion coefficient for the newly created gaps introduced by refinement. Wall heat transfer and boiling parameters were described using established empirical correlations. Ultimately, the subchannel code ATHAS-H was developed based on a homogeneous flow model with slip ratio. Model validation was conducted against a variety of flow and heat transfer experiments. For the PRIUS-Ⅱ experiment, compared to calculations from the CUPID code, the crossflow predicted by ATHAS-H captures the characteristics of the experimental data and demonstrates good agreement, indicating the refined subchannel model’s capability to represent crossflow distribution under inlet flow maldistribution in rod bundles. For the PSBT steady-state single subchannel benchmark, the void fractions predicted by ATHAS-H show good agreement with experimental values across four typical subchannel geometries, with most deviations within ±0.06, demonstrating its ability to accurately predict void fractions under typical PWR conditions. For CE5×5 rod bundle subcooled boiling tests, 95.02% of the calculated subchannel outlet temperatures deviate within ±5 ℃ of experimental values, and 95.8% of the calculated heated rod surface temperatures deviate within ±2.5 ℃. Compared to other codes based on traditional subchannel models, the composite refined subchannel model systematically improves prediction accuracy, confirming ATHAS-H’s capability in predicting subcooled boiling parameters and mitigating the channel box cold-wall effect under typical PWR conditions. Furthermore, computational time analysis for five typical CE5×5 cases indicates that ATHAS-H retains the computational efficiency characteristic of traditional subchannel codes. In conclusion, the composite refined subchannel code ATHAS-H is demonstrated to be suitable for predicting flow and heat transfer characteristics in rod bundle fuel assemblies. This study provides an analytical tool for predicting critical heat flux (CHF) and subcooled boiling in rod bundles under typical PWR conditions. Future work will include validation under more complex scenarios, such as transients and axially non-uniform power distributions, to further verify the code’s applicability and robustness. Additionally, to extend ATHAS-H’s application to more challenging conditions involving intense gas-liquid transients and complex two-phase flow patterns, its extension to a two-fluid model is deemed necessary.