Optimization of hydropower reservoir system using genetic algorithm for various climatic scenarios
Energy is an essential input for social and economic development. Due to the generalization of industrial and domestic activities, the energy demand has considerably increased. This causes a rapid growing in the level of greenhouse gas emissions and consequently increment in fuel prices. This princi...
Saved in:
Main Author: | |
---|---|
Format: | Thesis |
Language: | English |
Published: |
2015
|
Subjects: | |
Online Access: | http://psasir.upm.edu.my/id/eprint/71120/1/FK%202015%20131%20IR.pdf http://psasir.upm.edu.my/id/eprint/71120/ |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | Energy is an essential input for social and economic development. Due to the generalization of industrial and domestic activities, the energy demand has considerably increased. This causes a rapid growing in the level of greenhouse gas emissions and consequently increment in fuel prices. This principle was the driving force behind attempts to use clean and renewable energy sources such as hydropower. There are many reservoir systems around the world that have been constructed for hydropower generation. Also, hydropower provides a cheap source of electricity with less carbon emission. Although the renewable energy such as hydropower has obvious advantages, many of hydropower reservoir system are not operated efficiently and still being operated based on experience, rules of thumb or static rules appointed at the time of construction. It is noticeable that even small improvement in the operation rules can increase efficiency of a hydropower system. Accordingly, different operation policies were constructed and evaluated in this research. Generally, this research is divided into two main stages. The main scope of stage I is to maximize the power generation output by using the historical data (2003-2012). Accordingly, different forms of release policies, namely One Point Hedging Policy (1PHP), Two Point Hedging Policy (2PHP), Three Point Hedging Policy (3PHP), Discrete Hedging Policy (DHP), Standard Hedging Policy for Hydropower Generation (SHPHP), Binary Standard Operating Policy for Hydropower Generation (BSOPHP), and Standard Operating Policy for Hydropower Generation (SOPHP) were formulated and constructed using Matlab simulation. The developed models have been applied to the Cameron Highland and Batang Padang Hydro Scheme (CHBPHS) in Cameron Highland, Malaysia. CHBPHS is a cascade hydropower reservoir systems, which comprise of two reservoirs (Ringlet and Jor) and two power stations. In order to increase the system efficiency and maximize the power generation, constructed operation models were optimized. To determine the optimum solution in each policy, real coded genetic algorithm is used as an optimization technique. Thus, to enhance the functional efficiency in hydropower production, maximization of the total power generation over the operational periods is chosen as an objective function, while physical and operational limitations were satisfied. The results declared that by using the optimized hedging policies, the output of power generation could increase around 13% in the studied reservoir system compared to present operating policy (TNB operation). This considerable increase in power production will contribute in economic development. Moreover, the discrepancies of monthly mean power generation output between highest and lowest months by using hedging policies are around 10% in Ringlet reservoir and 26% in Jor reservoir, while this variation in power productions by TNB operation rules are about 30% and 49% respectively. Since hedging policies are usually applied to distribute the water supply, the power-supply also scatter in the simulation period. This is attributed to the effect of water distribution on power output. It can be concluded that these policies increase the stability of the system. The main scope of stage II is the prediction of future power generation by using generated weather data. Accordingly, the first aspect to point out is the generation of future climate parameters. Long Ashton Research Station-Weather Generator (LARS-WG) model is used firstly which was calibrated and validated using daily observed sunshine hours, rainfall, minimum and maximum temperature data. Afterwards, the minimum and maximum values of temperature and rainfall historical record were synthesize by the scenario file in order to predict the future climate parameters (Rainfall, minimum and maximum temperature) under possible scenarios. All scenarios reveal that climate change increases temperature around 0.3-0.7ºC at the location of the reservoir system. The increase in temperature could influence time and magnitude of rainfall by shifting dry and wet seasons. Moreover, the output results indicate a decrease in monthly rainfall. The output of LARS-WG model is used as an input of Artificial Neural Network (ANN). An ANN was subsequently applied as a rainfall-runoff modelling to predict the future stream flow feeding the reservoir systems. To explain more, ANN modelling comprised of two steps. The first step, ANN was calibrated and validated by using daily observed evapotranspiration, rainfall, and stream flow (2003-2012). In order to estimate daily evapotranspiration, daily observed Min and Max temperature was used in the estimation based on Hargreaves-Samani equation. By using the daily observed data, ANN can map the relationship between rainfall-runoff. The results indicate that the ANN model has good ability to capture the non-linearity of input/output in both training and test sets. In the second step, the future rainfall (output of LARS-WG) and future evapotranspiration (convert future minimum and maximum temperature generated by LARS-WG into future evapotranspiration by Hargreaves-Samani formula) are exported to ANN to predict the future stream flow under possible scenarios. After generating the future climate parameters, the predicted stream flow by ANN and estimated future evaporation (convert future minimum and maximum temperature generated by LARS-WG into future evaporation by penman formula) are exported to the constructed models to predict the future power generation output. The results declare that the future output of power generation will decrease under all possible climate scenario in both reservoir. According to the given results, the application of 3PHP for Ringlet reservoir and SHPHP policy for Jor reservoir, will give the highest amount of power that could be produced in the future and can be used to mitigate the negative effects of climate change. |
---|