Abstract: To enhance the mechanical properties of existing disassembled modular containerized structures and expand their application in barracks construction， a new structural system for containerized barracks has been developed based on traditional modular steel structure containerized house systems. This system is designed to be suitable for high-speed assembly and disassembly， offering greater flexibility in combination and ensuring floor insulation. The structural system comprises equilateral columns， thin-walled open-section beams， equilateral corner pieces， insulated floor systems， and foundations with adjustable heights. The introduction of this new structural system inevitably leads to changes in mechanical properties， making accurate calculation of structural responses is crucial. Based on the characteristics of disassembled modular container houses， this study employeds nonlinear structural analysis to establish three finite element models： beam element models， shell element models， and multiscale models for single and multi-story containerized units. It thoroughly analyzed the internal forces and deformation responses of components in single and multi-story units， quantified the differences in results caused by assumptions related to beam element models （rigid joints， pin joints） and shell element models （semi-rigidity）， and revealed the load transfer mechanisms resulting from interaction of main-secondary beams and structural configuration. Finally， it clarified the finite element models and refined analysis methods for containerized modular houses， providing valuable references for welding details between main and secondary beams for engineering design.Finite element comparative analysis showed that using beam element model hasd higher computational efficiency， but with lower accuracy of capacity and deformation which fell far short of requirements. The beam element model introduced the plane cross-section assumption automatically during analysis. Deformation results of such model could not reflect the contribution of twisting buckling of cold-formed section to deformation， nor could they reflect the interaction between main and secondary beams. Especially when predicting floor stiffness and deformation， it was not recommended to use beam element models and multiscale models for analysis. Under the "dead load + wind load" load case， the lateral displacement deviation of the multiscale model and the shell element model were marginal. Therefore， when evaluating the lateral displacement response of the structure， the multiscale model could be used for analysis. Multistory modular structural analysis could be carried out by using a hinged-beam element model for analysis， but the results of the beam element model were conservative： under lateral loading， there was a deviation of approximately 30% in deformation， and under vertical loading， there was a deviation of over 50% in internal forces. It was not recommended to use a rigidly connected-beam element model to analyze multistory modular structures， as the results of the rigidly connected-beam element model were comparably unsafe and with larger deviations compared to shell element models.