Cold thermal-energy storage technology for air conditioning is an important measure to achieve "peak shaving and valley filling". This study proposes a water-based cold thermal-energy storage solution that utilizes the space under metro-platform slabs to create a labyrinth-type cold water-storage tank. To investigate the feasibility of this solution, we first design a scaled-down model experiment based on the Reynolds-number similarity criterion and employ various turbulence models to numerically simulate the cold release and storage processes in a cold-water tank. The accuracies of these turbulence models for such mixed flows are analyzed by comparing the experimental and numerical results. The comparison shows that the large-eddy model is more accurate for modeling turbulent flow in this type of labyrinthine cold water-storage tank with mixed flows. Numerical simulations show that the efficiency of charging and discharging cold thermal energy using labyrinth-type water-based cold thermal storage in the space under metro-platform slabs is approximately 60%.