Integrating long- and short-term reservoir operation models via stochastic and deterministic optimization: Case study in Japan

Real-time reservoir operation models carried out under short operating horizons have no information about the seasonal changes in streamflow. The overall operation may improve provided that long-term information can be incorporated into the short-term process. This paper proposes a procedure that attempts to do such incorporation by means of stochastic and deterministic optimization. Long-term (monthly) and short-term (daily) reservoir inflow forecasts are assumed to be uncertain and reliable, respectively. The idea is to first solve a monthly operation model by an explicit stochastic programming approach and then use its information to guide the daily operation, which is solved by deterministic optimization. The monthly information will decide whether hedging should start being implemented during the daily operation in order to mitigate a potential future shortage. This stochastic-deterministic procedure is applied to a case study in Japan and the results show that it provides operating rules relatively similar to the ones obtained when perfect forecasts are used for the monthly inflows.