<p>The on-orbit assembly of large and flexible structures is a critical capability for future space missions, yet poses significant challenges in modeling and control dynamics. Traditional approaches often fail to capture the dynamic coupling between a robotic assembler and the evolving structure, especially under varying inertia and flexibility conditions. This paper addresses this gap by presenting a modeling framework tailored for the preliminary design phase of on-orbit assembly missions. This work presents a methodology for modeling an on-orbit assembly scenario involving a large flexible structure constructed by a multi-arm robotic system. A three-legged walking robot is responsible for building the structure, with its primary goal being to walk stably on the flexible structure while picking up, carrying and assembling substructure components. The proposed model, based on Linear Fractional Representations (LFR), captures the dynamic interactions among subsystems, incorporating changes in geometry, flexible dynamics, and parametric uncertainties. Although the model is limited to linear behavior, it enables structured analysis for robust control design and performance evaluation. A frequency-domain path optimization strategy is also introduced to select assembly trajectories using cost functions related to stability and performance objectives. Results demonstrate the effectiveness of both the modeling framework and the optimization approach in supporting the design and validation of on-orbit assembly missions.</p>