Abstract: The double-tube once-through steam generator (DOTSG) has significant advantages, including high heat exchange efficiency, a compact structural footprint, and a small overall size. These advantages have established its widespread application within the field of small nuclear reactors. In this paper, in response to the challenge in predicting the three-dimensional thermal-hydraulic characteristics of DOTSG caused by complex flow channel geometry, a novel multi-scale coupled modeling framework was proposed. This framework was designed to integrate a macroscopic representation of the entire tube bundle system with a mesoscopic analysis of the detailed flow channels. Based on this foundational framework, a three-dimensional thermal-hydraulic analysis code DOTAF was developed, utilizing a porous media model as its core computational approach. Within the DOTAF code, a porous media model was implemented on the primary side to efficiently simulate the three-dimensional flow patterns and heat transfer processes throughout the tube bundle region. The Chexal-Lellouche drift-flux model, which was recognized for its suitability in modeling narrow-channel configurations, was implemented on the secondary side characterized by narrow annular flow channels to accurately capture the two-phase flow and phase change processes. To ensure a cohesive and high-fidelity simulation, the parameters between the primary and secondary sides were coupled in real time through a dynamic bidirectional grid-node data mapping algorithm, which facilitates continuous and accurate data exchange across the computational domains. The reliability and predictive accuracy of the DOTAF code were subjected to rigorous validation. This validation was conducted against experimental data obtained from the ACP100 reactor prototype. The results demonstrate strong performance across the full spectrum of operational conditions. Specifically, for all key thermal-hydraulic parameters at the outlets of both the primary and secondary sides, the average prediction error is maintained below 3%. Furthermore, the maximum recorded error in any test case remains under 5%, confirming the code’s robustness. Beyond validation, the DOTAF code was deployed for a comprehensive numerical simulation study focusing on a specific DOTSG designed for an integral inherent safety reactor, which successfully demonstrates the program’s capability. DOTAF proves effective in revealing the complex, inherent three-dimensional thermal-hydraulic behaviors within the DOTSG and in elucidating the key underlying physical mechanisms governing its performance. In summary, this research project has culminated in the development of the DOTAF analysis tool. This work provides an efficient, reliable, and three-dimensional analytical resource. The tool is directly applicable to supporting several critical engineering activities for compact steam generators, including their refined design, comprehensive safety assessment, and operational optimization.