How Much Information on Future Streamflows is Utilizable in Design and Operation of Dam Reservoirs?

The problem of analyzing the storage-yield-reliability of surface water reservoirs is a classical problem in water resources engineering. This study explores the effect of the horizon of foresight on future inflows on the minimum reservoir capacity required for meeting a specified yield. The problem was formulated as mixed integer linear programs (MILP). We analyzed the role of perfect foresight assumption on future inflows by formulating the problem as a series of finite-horizon mixed integer linear programs (MILPs), each with a limited, partial foresight. The MILPs were then solved sequentially over a long-run planning horizon for determining the minimum reservoir capacity required for meeting a specified reservoir yield at a certain level of reliability. Although it was generally true that the required reservoir capacity will increase if the foresight on future inflows becomes more limited, the study showed that an opposite result was also possible, depending on the characteristics and hydrology of the system. Moreover, after a certain level of foresight horizon, knowing about the future inflows does not help to get a lower required reservoir capacity. These findings provide insight into the issue of uncertainty of future inflows and their forecasts in design and operation of water reservoir systems.