Optimization of operational policies for the Minab Reservoir, Southern Iran

Water resources management in an arid region with severe drought, such as the Minab area, in Iran, is very critical. Periods of drought have resulted in severe stress on the amount of current inflow to the reservoir. These severe stresses could not be addressed by the reservoir management through the standard operation policy method. The operation policy should be changed to consider the drought period conditions. There is a need for a reservoir management system to optimize water allocation policies during the inadequate water supply periods. Forecasting and accurate estimation of the future water inflow to the reservoir are the most important challenges in the management of water resources for the system. In the past, the management of the reservoir was widely concentrated on developing the operating rules in managing water resources. This research focuses on the optimization policies combined with a forecasting model for reservoir operation during drought conditions. The main objective of this research is to develop and optimize the operational policy for managing the Minab reservoir operation to maximize the benefit of water release from the reservoir for different demand scenarios. The Soil and Water Assessment Tools (SWAT) model and the Focused Time Delay Recurrent Neural Network (F.T.D.N.N) method were used to simulate and forecast future inflow to the reservoir by considering stream flow factors and their constraints. The FTDNN was found to produce a higher accuracy and thus was selected as the forecasting model. The current operation model uses the standard operation policy (SOP) rules to simulate the water demand and to recognize the shortage of agricultural water demand as the main demand sector. The SOP estimated the shortage of water for agricultural allocation in a monthly, three- month, six- month or yearly periods. To solve the shortage of water during drought, the Limiting Hedging Rules model and Genetic Algorithm (GA) were developed to determine the optimal allocation for agricultural demand. Through the hedging rule optimization an algorithm was developed to determine the benefit of water release and the water conserved in the reservoir. Also, three triggers were estimated for use as guidelines for managing the reservoir during drought occurrence. To mitigate the drought condition the probable scenarios, policies and management were applied. Rule curve for five possible scenarios were optimized by using Genetic Algorithms. The agricultural management optimization was applied to optimize the parameters like area, relative yield water requirements and irrigation efficiency. These parameters were optimized to reduce the water requirement based on the cost and benefit by using the Lingo model. As a result, when severe drought occurs, using optimized operational policies combined with the forecasting model could have a significant effect on reducing drought severity in an arid region. This research shows that a combination of forecasting and optimized operational policies models can be applied to manage drought conditions in water resource management. An algorithm was also developed for reservoir storage to determine the benefits of water conservation and reservoir water release to optimize reservoir operations in drought periods. Comparing the performance of the different optimized operational policy models showed that the Hedging Rule, SOP, and Genetic Algorithms (GA) can respectively allocate 2404, 1991, 1811 million cubic meters (MCM) of water for agriculture during the shortage period while the benefit values of following these models were estimated to be 10,915, 7,395 and 6,075 thousand US dollars respectively.

Preview PDF FK 2012 119R.pdf Download (753kB) | Preview

Actions (login required)

View Item