Comparison of Dynamic, linear and Nonlinear Programming Approaches in Optimal Operation of Reservoir

Application of efficient policies to operate dam reservoirs is of great importance due to water shortages and frequent droughts in the country. For this، different approaches can be implemented، which are based on different modeling systems. The present paper aims to compare some of these approaches for optimal water allocation in Zayandeh Rud agriculthral system during 1998-2000 droughts using some evaluation criterion. The mentioned approaches include: 1) dynamic programming (DP) considering constant agricultural water demand and no uncertainty in inflows، 2) stochastic DP (SDP) with constant agricultural demand and uncertainty in inflows and 3) Sampling SDP (SSDP) considering constant demand and inflow uncertainties and 4) variable agricultural water demand with respect to crop types، growing stages، irrigation systems using linear and non-linear programming (LP and NLP) modeling and no uncertainty in inflows. Results showed the first and fourth approaches had the simplest and most complicated modeling process، respectively. However، with respect to some efficiency indices like temporal and volumetric reliability، maximum as well as total shortages، the stochastic approaches i. e. the second and third ones outperformed respectively equal to 38%، 68%، 157 and 330 MCM. This superiority was considerable especially for the third modeling approach. The comparison of net benefit values from optimization approaches as well as real obtained one for Zayande Rud system showed the optimization approach could increase the benefit from 68% to 73%.