Reservoir system modelling using nondominated sorting genetic algorithm in the framework of climate change
Reservoir systems require a continuous development of models for optimal operations in the context of future climate change. The statistics of climate change variation and the growth of evolution method inspire the development of a sustainable and long term reservoir operation system. The aim of the...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2014
|
| Subjects: | |
| Online Access: | http://umpir.ump.edu.my/id/eprint/10690/ http://umpir.ump.edu.my/id/eprint/10690/ http://umpir.ump.edu.my/id/eprint/10690/1/Civil%20-%20nurulnadrahaqilahbintitukimatpa%20%28CD8350%29.pdf |
| Summary: | Reservoir systems require a continuous development of models for optimal operations in the context of future climate change. The statistics of climate change variation and the growth of evolution method inspire the development of a sustainable and long term reservoir operation system. The aim of the study is to propose and establish a sustainable and long term reservoir operation system and management with adaptation to climate change using the integrated models. The models are Nondominated Sorting Genetic Algorithm type II (NSGA-II), Linear Programming (LP), Statistical Downscaling Model (SDSM) with Multi-linear Correlation Matrix (M-CM), hydrological and crop models. There are two groups of model proposed in this study namely as Model A and Model B. Model A is a combination methods of SDSM with M-CM, hydrological and crop model which considered the climate change variation. Whilst, Model B exclude the climate change factor by using Valencia Schaake model (VS) and Thomas Fiering model (TF). The reliability, resiliency, and vulnerability of models were evaluated. These models were tested at Pedu-Muda reservoir system that supplies water for the Muda Irrigation Scheme, Kedah, Malaysia. The inclusion of M-CM as a screening tool to the SDSM model was successful to produce lower mean absolute error (MAE = 4mm/day), mean square error (MSE = 29mm/day), and standard deviation (St.D = 1mm/day). It is expected that the future rainfall and temperature would increase about 4% and 0.2oC per decade respectively. The volume of paddy water requirement is expected to reduce 0.9% per decade. This is due to the increment of rainfall and uncontrolled flow in the paddy field. The generated synthetic inflow series produced +0.4% and -1.3% discrepancies from the historical records using VS and TF models respectively. The NSGA-II and LP models successfully established a sustainable reservoir operation for long term period. Furthermore, the NSGA-II was successful in satisfying the multiple objectives demand and provides a set of alternative solutions that were presented in the Pareto optimal curves depending on the climate pattern. The rule curve formation for Model A was consistently higher than Model B in range of 1% to 5%. In terms of reliability, resiliency, and vulnerability evaluation, the Model A-NSGA-II is remarkably and potentially promised sufficient water supply throughout the year. Model B (VS) was the second best performer followed by Model B (TF). In conclusion, this finding contributes toward the development of models using evolution algorithm and statistical methods for sustainable water resources planning and management in the context of future climate change |
|---|