A Data-Driven Traffic Assignment Framework for Mixed Lane-Free Urban Networks

Mixed and weakly lane-disciplined traffic systems pose a significant challenge for data-driven transportation analysis, as conventional lane-based models fail to capture heterogeneous space usage and congestion dynamics. This limitation is particularly acute in cities where high-resolution traffic data increasingly reveal complex, non-lane-based vehicle interactions that are poorly represented by standard analytical tools. This paper proposes a network-level simulation framework for mixed, lane-free urban traffic based on the concept of area occupancy, which quantifies congestion through collective road space usage rather than lane-based density. The framework integrates class-specific vehicle characteristics, intersection-level space constraints, and network-wide flow propagation within a scalable computational structure. This is particularly useful in multi-class environments where different vehicle types may experience and contribute to congestion differently. The model was first validated on two hypothetical networks with one-way and two way links to ensure proper flow propagation and flow conservation within the network. To demonstrate scalability and robustness, the framework was applied to the Sioux Falls network. To demonstrate model’s potential for real-world traffic assignment in complex urban environments, the framework is applied to a selected network in Jodhpur City, India. By bridging traffic flow modelling with data-oriented simulation, this study provides a methodological foundation for more realistic congestion analysis and traffic assignment in mixed traffic environments.