| Author | Le Bat |
| Call Number | AIT Diss. no.WA-81-01 |
| Subject(s) | Water resources development--Mathematical models
|
| Note | A Dissertation submitted in partial fulfilment of the requirement for the Degree of Doctor of Engineering, School of Engineering and Technology |
| Publisher | Asian Institute of Technology |
| Series Statement | Dissertation ; no. WA-81-01 |
| Abstract | The present study analyses the optimization of the operation of multi-purpose multi-reservoir water resources systems, which explicitly takes into account the stochastic nature of monthly streamflows. In order to overcome the difficulties in the conventional dynamic programming, two optimization models with theoretical backgrounds based on differential dynamic programming have been proposed in the study. The first model, called the stochastic discrete differential dynamic programming (SDDDP) model is developed with the assumption that the monthly streamflows in the systems are independently distributed. Normal and log-normal distributions are used to fit the monthly streamflows. As a consequence, the objective function in terms of expected benefits in the model can be analytically derived and computed without discretizing the distribution of streamflows. Reliability constraints are introduced into the formulation of the model as normal constraints without using simulation runs. The second model, called the stochastic constrained differential dynamic programming (SCDDP) model, takes into account the serial correlation of monthly streamflows by using the conditional and transitional probability matrices. The objective function in terms of expected benefits in the model is approximated by a quadratic function. As a result, a quadratic programming algorithm instead of enumeration methods is used in the optimization procedure at each stage of the operation cycle. To demonstrate the applicability of the two proposed models, the Mekong river system, which is composed of three serially linked reservoirs which are used for the purposes of hydropower generation, irrigation and flood control, was used as the case study for the application of the proposed models. A one-by-one optimization procedure was developed for the efficient generation of the initial trial trajectory in the SDDDP model. A study on the selection of the appropriate number of inflow intervals for the discretization of the distribution of streamflows was carried out in the SCDDP model. Lastly, the performance of the optimal solutions obtained from the two models was studied and compared using a simulation procedure. The SCDDP model produced a more flexible operating policy than the SDDDP model. However, the former model required a larger computer memory. The results obtained from the two models were very close to those yielded from the simulation tests using historical record of streamflows. |
| Year | 1981 |
| Corresponding Series Added Entry | Asian Institute of Technology. Dissertation ; no. WA-81-01 |
| Type | Dissertation |
| School | School of Engineering and Technology |
| Department | Department of Civil and Infrastucture Engineering (DCIE) |
| Academic Program/FoS | Water Resources Research Engineering (WA) |
| Chairperson(s) | Selvalingam, Selvadore; |
| Examination Committee(s) | Anat Arbhabhirama;Suphat Vongvisessomjai;Gupta, Ashim Das;Huynh Ngoc Phien; |
| Scholarship Donor(s) | Mekong Committee; |
| Degree | Thesis (Ph.D.) - Asian Institute of Technology, 1981 |