Mind Over Model: Optimization of Gate Closure for the South Hartford Conveyance and Storage Tunnel

Hydraulic modeling is frequently used for planning and design of deep tunnel conveyance and storage systems. However, care must be taken to check that the model results are reasonable and that the end result is the best possible design. This paper illustrates some examples of how human “thinking” can and should be used when reviewing model results. These principles are illustrated using the design of the South Hartford Conveyance and Storage Tunnel (SHCST) in Hartford, CT. This project includes a South Tunnel and a North Tunnel. The South Tunnel is under construction while the North Tunnel is currently still in the planning phase. However, the South Tunnel must work as part of an integrated system with the North Tunnel, and therefore the analyses described in this paper included both tunnels. During wet weather, CSO / SSO flows will be diverted into the tunnel when the capacity of the collection system is exceeded. Hydraulic modeling was used to design and optimize the control logic for the gates that regulate flows into the tunnel. Hand calculations of the initial model results indicated discrepancies which were resolved by decreasing the length of the tunnel segments and time step used in the model. The model was subsequently applied to determine the optimal gate closure based on 54 year continuous simulations. Hydraulic modelers often try to make the model completely self-contained, so that the gates in the model automatically open and close based on calculated parameters in the model, such as the hydraulic grade line (HGL). Analysis of the model results showed that extreme storms in the period of record that caused the tunnel to fill would have been predicted by modern weather forecasting days in advance. As a result, human “Predictive Control” could be used to further optimize the operation of the tunnel by closing the CSO gates earlier than the pre-programmed set-points used in the hydraulic model. Another aspect of the tunnel design that involved imposing “human thinking” over raw model results was sediment calculations. Sediment modeling was performed to estimate the distribution and quantity of sediment in the tunnel over time. There is a lot of uncertainty in sediment modeling, and therefore the results were checked in multiple ways, including using two different sediment models. Through careful thinking about the model results and how the tunnel will actually be operated, an optimal tunnel size and operating strategy was developed for the SHCST.