Urban systems complexity requires advanced modeling approaches for effective city management and planning. This study develops comprehensive digital twin frameworks for smart cities, integrating real-time data from IoT sensors, traffic systems, energy grids, and social media feeds. We created digital replicas of three urban districts across Cairo, Dublin, and Taipei, processing 2.4 TB of data daily through machine learning algorithms for predictive modeling. The systems successfully predicted traffic congestion with 91% accuracy, optimized energy distribution reducing consumption by 18%, and identified urban heat islands enabling targeted interventions. Real-time simulation capabilities allowed testing of urban policy scenarios before implementation. The research provides scalable architectures for digital twin deployment in diverse urban contexts.