Optimum reservoir operation of a networking reservoirs system using conditional atom search optimization and a conditional genetic algorithm.

This study aimed to apply conditional atom search optimization (CASO) for searching optimum rule curves in a networking reservoirs system with a reservoir simulation model. The networking reservoirs system consisted of 5 reservoirs located in Sakon Nakhon Province, Thailand. The efficiency of the new optimum rule curves was determined by comparison of operating systems between a single reservoir and a networking reservoirs system. The results displayed circumstances of scarcity and excess of water. Where the circumstances of scarcity are frequency and duration. Whilst, excesses of water are average water and the highest water. In addition, the efficiency of searching for optimum rule curves was compared between conditional genetic algorithm (CGA) and CASO techniques. The new optimum rule curves from the networking reservoirs system had an average excess water of 43.828 MCM/year. This average excess water was less than that found for optimum curves from the single system in which the average excess of water was 45.602 MCM/year. CASO was more efficient in converging optimum rule curve solutions faster than CGA by 40.00%. In conclusion, the CASO can be used to search for optimum networking reservoirs rule curve solutions effectively. For the networking reservoirs system derived water from the upstream reservoirs, an analysis was performed of the downstream reservoir. The results showed that the optimum rule curves using CASO operated as a networking reservoirs system provided higher efficiency than a single reservoir system. In addition, they reduced the amount of time that water exceeded the river capacity at a downstream weir by one month compared with the original period of two months.