Urban flooding remains a critical challenge worldwide, driven by rapid urbanisation and climate change, often resulting in significant economic damage and loss of life. While traditional hydrodynamic models based on shallow water equations (SWEs) offer high accuracy, they are computationally intensive and inefficient for large-scale, real-time applications. Conceptual models are computationally efficient but often compromise accuracy, especially in capturing inertial effects. This study introduces a novel Cellular Automata-based Dynamic Fast Flood Model (CADFFM) that incorporates Bernoulli hydraulic head to account for kinetic energy and better replicate complex hydrodynamic phenomena such as backwater effects, hydraulic jumps, and transcritical flows. The model is coupled with the 1D EPA SWMM to form Hybrid CADFFM, enabling bidirectional interactions between surface and subsurface drainage networks. The model is validated using three benchmark dam-break experiments and applied to a real-world urban catchment, where it demonstrates reliable accuracy in predicting flood extents and maximum depths, and performs simulations approximately twice as fast as MIKE FLOOD. These findings underscore the model’s potential as a computationally efficient and robust tool for urban flood prediction and management.