Author(s): Rosalva Mendoza Ramirez

Linked Author(s): Rosalva Mendoza Ramírez

Keywords: Stochastic Dynamic Programming; Operational policies; Multipurpose reservoir

Abstract: The work presents operational policies for the water of Cointzio dam, located 13 km south of the city of Morelia, in Michoacán, Mexico. This reservoir was opened in the early 1940s, and supplies water continuously to the purification plant at Vista Bella, which provides water to the south of city. It also feeds 4 of the 5 modules for irrigation in Irrigation District 020. A streamflow record (1940 – 2016) was used to model the amount of water to be extracted from the Cointzio reservoir in monthly time intervals. Stochastic dynamic programming was used and two discretization intervals were used for the useful capacity of the dam. The months of the year were grouped into two stages. The probabilities associated with the input volumes were calculated and maximum and minimum extractions were defined according to the monthly water volumes required. Seven combinations of penalization for spill and deficit events were also defined. Based on results of the spill, deficit and minimum storage variables for the period, the best discretization interval and the two best operational policies were selected. The results show that the requirements of the water purification plant are almost always met. However, the demands of the irrigation modules are less often met. The operational policies obtained are easy to use, as they are presented as a state-extraction matrix. The very short running time (seconds) is a further advantage, and operators can set several combinations of penalization for deficit and spill events which allows new rules to be tested.

DOI: https://doi.org/10.3850/IAHR-39WC25217119202272

Year: 2022