Hybrid concentrating solar power systems integrated with high-temperature steam electrolysis provide the concurrent generation of electricity, heat, and hydrogen with elevated overall efficiency. The real-time simulation and supervisory control of integrated thermal-hydraulic networks are limited by the lack of fluid-flexible solvers and hierarchical control frameworks that can manage multi-regime thermodynamic behavior and mode switching. This study introduces a GPU-accelerated, generic, object-oriented solver-controller framework that integrates modular component representations with a hierarchical supervisory control architecture, facilitating precise dynamic modeling and stable control of hybrid thermal-hydraulic systems. A backward-differentiation thermal-hydraulic network solver is integrated with modular component representations and a three-layer hierarchical control structure, which features a classifier-based supervisory layer for steam-source selection and adaptive regulators for load distribution, thermal storage interaction, and electrolysis thermal management. The suggested algorithm enables district heating network simulations to achieve a 40-50 % reduction in execution time under worst-case scenarios when executed on a GPU, facilitating real-time performance. The framework replicates industrial reference results with deviations under sub-percent, ensures stable operation during 45 % load ramps, restricts solar steam generator pressure variations to +/- 2 bar, maintains electrolysis thermal gradients below 10 K/cm, and produces hydrogen with a mole fraction of 0.82 at 80 % steam utilization. The suggested modeling and control framework offers a transferable foundation for the real-time development of digital twins and supervisory control of hybrid thermal-hydraulic energy systems, establishing a platform conducive to the future incorporation of optimization strategies and artificial intelligence-driven acceleration.