Extraction of Static and Dynamic Reservoir Operation Rules by Genetic Programming

Abstract

Considering the necessity of desirable operation of limited water resources and assuming the significant role of dams in controlling and consuming the surface waters, highlights the advantageous of suitable operation rules for optimal and sustainable operation of dams. This study investigates the hydroelectric supply of a one-reservoir system of Karoon3 using nonlinear programming (NLP), genetic algorithm (GA), genetic programming (GP) and fixed length gen GP (FLGGP) in real-time operation of dam considering two approaches of static and dynamic operation rules. In static operation rule, only one rule curve is extracted for all months in a year whereas in dynamic operation rule, monthly rule curves (12 rules) are extracted for each month of a year. In addition, nonlinear decision rule (NLDR) curves are considered, and the total deficiency function as the target (objective) function have been used for evaluating the performance of each method and approach. Results show appropriate efficiency of GP and FLGGP methods in extracting operation rules in both approaches. Superiority of these methods to operation methods yielded by GA and NLP is 5%. Moreover, according to the results, it can be remarked that, FLGGP method is an alternative for GP method, whereas the GP method cannot be used due to its limitations. Comparison of two approaches of static and dynamic operation rules demonstrated the superiority of dynamic operation rule to static operation rule (about 10%) and therefore this method has more capabilities in real-time operation of the reservoirs systems.

Keywords

• NLP
• GA
• GP
• Fixed length gene GP (FLGGP)
• Static and dynamic operation rules