Supercritical Stormwater Flow Interception through Bottom Rack with Transverse Barrier

The Hong Kong West Drainage Tunnel (HKWDT) system consists of 34 stormwater intake structures designed to intercept stormwater runoff in upland catchments. During stormwater events, each of these intake structures intercepts and diverts the upstream supercritical flow into a bottom rack chamber connected to a supercritical vortex drop. Small surface flows are conveyed to the downstream drainage system through a narrow low-flow channel (LFC) along one side of the intake. The LFC is designed for a capacity of $20\mathrm{L}/\mathrm{s}$. However, when the intakes are in operation, observations suggest that the stormwater flows conveyed via the LFC might be significantly more than the design values. We report an investigative physical and numerical modeling study of the hydraulic performance of a representative existing intake structure. A 1:12 undistorted physical model based on Froude similitude was designed to study the complex hydraulic interaction of the incoming supercritical flow with the transverse barrier, bottom racks on top of the rack chamber, and a low-flow channel in supercritical flow regime. The interception capacity and flow characteristics of the bottom rack intake is predicted via a two-dimensional (2D) numerical solution of the shallow water equations using a shock-capturing finite-volume method. Physical model tests and numerical experiments were carried out for representative design flow events including those for storms with 2-, 10-, 50-, and 200-year return periods with climate change. The predicted flow features—including the discharge intercepted by the bottom rack chamber and the low-flow channel—were compared with experimental observations. The results indicate that the unintercepted flow in typical rainstorm event was $0.5\mathrm{L}/\mathrm{s}$ in the physical model, corresponding to $0.25{\mathrm{m}}^3/\mathrm{s}$ in field conditions. Under a 200-year rainstorm with climate change, the unintercepted flow was $2.3\mathrm{L}/\mathrm{s}$ in the physical model, corresponding to $1.15{\mathrm{m}}^3/\mathrm{s}$ in field conditions. Overall the model results show that around 10% of the inflow enters the LFC at high inflows—around $2\mathrm{L}/\mathrm{s}$ or $1{\mathrm{m}}^3/\mathrm{s}$ of prototype flow—which is much higher than the $20\text{-}\mathrm{L}/\mathrm{s}$ design target; this confirms the observation that the stormwater runoff conveyed to the downstream drainage system is more than originally designed. Finally, through a combination of heuristic reasoning and physical model studies, the existing intake structure design was modified to improve interception performance. The improved interception was verified through hydraulic model simulation. Furthermore, the applicability of spatially varying rack discharge coefficient for supercritical flow above bottom racks in a 2D high-resolution hydraulic numerical model was validated.