Optimal location of urban land-uses considering transportation systems using genetic algorithm

The optimal spatial allocation and determination of the appropriate number of land-uses have received limited attention in urban transportation planning, often leading to increased congestion and substantial costs within transportation systems. To address this gap, this study introduces a bi-level optimization model designed to minimize transportation costs while simultaneously identifying both the optimal location and number of main land-uses. The upper level aims to reduce total system travel costs, whereas the lower level simulates user equilibrium behavior under varying land-use distributions. A Genetic Algorithm (GA) is employed to solve the model. The proposed approach is applied to the Sioux Falls Network as a numerical case study. Results demonstrate the model’s effectiveness in significantly reducing transportation costs compared to traditional monocentric and polycentric (non-optimized) approaches. For instance, in scenarios with 100 land-uses, total transportation costs decreased from 159.84 million (monocentric) and 56.89 million (standard polycentric) to 14.53 million using the proposed model. Furthermore, convergence analysis confirms the robustness of the GA, showing rapid cost reduction during early iterations and stable attainment of a near-optimal solution. Overall, the findings highlight that decentralization, when integrated with transportation system optimization, can yield substantial and sustainable improvements in urban efficiency. These results underscore the critical importance of incorporating transportation performance into land-use planning to achieve long-term cost optimization and network resilience.