Reservoir Operation using Combined Genetic Algorithm & Dynamic Programming for Ukai Reservoir Project

D.K.Gajjar, MSU, Baroda, Gujarat, India; P.V.Shah ,IIET, Dharmaj, Gujarat, India

Dynamic Programming, Genetic Algorithm, Reservoir Operation, Ukai Reservoir

Operation of reservoirs, often for conflicting purposes, is a difficult task. The uncertainty associated with reservoir operations is further increased due to the on-going hydrological impacts of climate change. Therefore, various artificial intelligence techniques such as genetic algorithms, ant-colony optimization, fuzzy logic and mathematical optimization methods such as Linear programming, Dynamic Programming are increasingly being employed to solve multi-reservoir operation problems. For doing optimization, objective function is formulated which is subjected to various constraints. Constraints include continuity equation, reservoir storage constraints, release constraint and overflow constraint. Monthly data for the study are used of year 2007 to 2011. Genetic algorithm is based on Darwin’s theory of Survival of the fittest. GA reduces the difference between releases and demand and returns the value of the fitness function / Objective function. In 2007, using Genetic Algorithm the generation of power can be increased 9.22% through optimal releases. There is 7.14% increase in optimal reservoir release. Further include the study of to optimize the monthly releases from the reservoir i.e. to minimize the sum of the squared difference between monthly release of water from the reservoir and downstream demands for Ukai Reservoir Project. DP is a quantitative technique which converts one big/large problem having many decision variables into a sequence of problem each with a small number of decision variables. DP reduces the difference between releases and demand and returns the value of the Objective function. After that the difference between actual releases and optimal releases i.e. Maximum Absolute Error is calculated for a month of July, August, September and October for each year. Also for evaluation of models developed by using dynamic programming the Root Mean Square Error and Correlation coefficient is calculated for all models. And also net additional available water for every year is also carried out.
    [1] Ashok, K. (1999) Application of Genetic Algorithms for Optimal Reservoir Operation. M.Tech thesis, IIT, Kharagpur, India. [2] Goldberg, D.E. and Deb, K. (1990). A comparative analysis of selection schemes used in genetic algorithms. Foundations of Genetic Algorithms, ed. G.E. Rawlins, Morgan Kaufman, San Mateo, Calif. pp. 63–93. [3] Reddy L.S. (1996) Optimal Land Grading Based on Genetic Algorithms. J. of Irrigation and Drainage Engineering ASCE, 122(4), pp. 183-188. [4] Elmahdi,A., Malano, H. and khan, S.(2004) “A system dynamic approach and irrigation demand management Modelling”, Environmental Engineering Research Event 2004 conference 6-9 December 2004. Published by University of Wollongong Press ISBN: 1 74128 080 X. (www.ere.org.au). [5] Yeh, C.H., Labadie, J.W. (1997) “Multiobjective watershed-level planning of storm-water detention systems”. Journal of Water Resource Planning Management ASCE 123:336–343 [6] Yang, C, Chang, L. Yeh, C. and Chen, C. (2007). Multiobjective planning of surface water resources by multiobjective genetic algorithm with constrained differential dynamic programming. Journal of Water Resources Planning and Management. 133(6), 499- 508
Paper ID: GRDCF001015
Published in: Conference : Recent Advances in Civil Engineering for Global Sustainability (RACEGS-2016)
Page(s): 71 - 77