1
Operation policy for a complex water resources system on the basis of stochastic dynamic programming | |
Author | Kularathna, Maddumage Don Udayananda Pushpakumara |
Call Number | AIT Thesis no. WA-88-8 |
Subject(s) | Water resources development--Systems engineering |
Note | A thesis submitted in partial fulfilment of the requirement for the degree of Master of Engineering, School of Engineering and Technology |
Publisher | Asian Institute of Technology |
Abstract | Application of Stochastic Dynamic Programming (SDP) for the derivation of optimal operation policies for multireservoir systems is limited by the dimensionality problems inherent to SDP. Operation policies for two serially l inked reservoir subsystem in the Mahaweli Development scheme of Sri Lanka have been derived by NANDALAL( 1986) using a conventional SDP model. The intention of selecting only two reservoirs f rom the above multireservoir system has been to limit the huge computational and computer memory requirements of SDP algorithm to an affordable level . This study extends the analysis to include a third reservoir as well in the s ubsystem under consideration. Four different SDP based techniques are tested for their applicability to derive optimal operation policies for the system under consideration. The first technique is a sequent optimization approach which optimizes the cascade reservoir system starting from the uppermost reservoir, and proceeding downwards. Second approach considers the two downstream reservoirs as a single hypothetical Composite reservoir. An iterative SOP/simulation based approach is introduced as the third technique. A conventional SOP algorithm is used as the fourth approach, which considers the three-reservoirs-in series configuration in the optimization process . The operation policies derived using SDP based mode l s are tested using simulations according to the derived policies. The results of the simulations are compared with the deterministic optimum operation of the system obtained by an Incremental Dynamic Programming (IDP) mode l. Although not applicable in general , the sequent optimization ha s been the best approach to derive operation policies for this reservoir system. The suitability of a Composite representation of Victoria and Randenigala reservoirs has been justified. The iterative approach; although provided good results, has consumed a comparatively large computer time. Operation policy derived by the three-reservoir-model cannot be used as an adequate guide line for the actual operation. This is due to the fact that a wide range of inflows and storages have been represented by only a few representative values, in order to limit the computational and memory requirements . |
Year | 1988 |
Type | Thesis |
School | School of Engineering and Technology (SET) |
Department | Other Field of Studies (No Department) |
Academic Program/FoS | Water Resources Research Engineering (WA) |
Chairperson(s) | Bogardi, Janos J. |
Examination Committee(s) | Gupta, Ashim Das ;Nielsen, Steen Asger |
Scholarship Donor(s) | Government of Netherlands |
Degree | Thesis (M.Eng.) - Asian Institute of Technology, 1988 |