Effective flood mitigation in urban areas requires efficient modelling tools to evaluate surface-drainage flow dynamics under various rainfall scenarios. Coupled one-dimensional-two-dimensional (1D-2D) models are increasingly used for Stormwater Drainage (SWD) design and performance evaluation. However, their city-scale application is constrained by high computational demands arising from fine-mesh requirements, ground elevation inconsistencies across models, and numerical instability, especially near flow interaction junctions. To address these challenges, the study introduces a structured framework to generate a node-concentrated mesh that explicitly uses the SWD network topology by selectively constraining and refining the mesh around drainage junctions where flow exchanges occur, rather than relying on conventional geometry-based refinement. This method provides two key advantages: (i) improves ground elevation consistency between surface terrain and drainage nodes at junctions while minimizing model coupling discrepancies, and (ii) optimizes the trade-off between computational cost and accuracy. The proposed approach is validated within a SWMM-MIKE + coupled 1D-2D modelling framework for historical flood events in a 27.8 km2 urban catchment in Chennai, India. Synthetic 2-, 5-, and 10-year design storms are then simulated to assess SWD performance. An Integrated Drainage Efficiency Index (IDEI) is introduced, based on spatially distributed reductions in simulated flood extent, area-weighted flood depth, and area-weighted duration, providing an interpretable index that evaluates the overall effectiveness of the SWD system in mitigating urban flooding under different rainfall conditions. Overall, the mesh-generating method and the modelling framework offer a practical tool for analysing large-scale urban floods, supporting more informed and resilient drainage planning in rapidly growing cities.