Empirical Analysis of Chilled Water Generation for Off Peak Period of Cogeneration plant Using Neural Network

摘要

Cogeneration plant has been widely used in Malaysia to produce electricity and chilled water. However, during the off peak hour most of the cogeneration plants in Malaysia operate to produce only electricity and dissipated waste heat to the environment which increases greenhouse gas emissions, and contributes more to global warming. Thus, this study focuses to evaluate the amount of a waste heat generated at off peak hour using the peak hour waste heat generated data. Neural network (NN) was utilized to study the trend of the off-peak hour data to estimate the possible amount of chilled water could be generated instead of releasing it to the environment. Forty-three weeks of cogeneration plant data of year 2011 were taken to develop the train model between waste heat and chilled water using the peak hour data. The trained model was tested and varied using means square error and R values. The minimum R values is 0.91 which means the developed train model can be used to estimate the possible amount of chilled water during off peak hour. The result shows that there could be a possibility to generate 121464.93 RTh chilled water on average weekly during off peak period. This amount is 76% of the amount of weekly generated during peak hour. Based on the analysis, if the cogeneration plant utilizes the waste heat generated during the off peak hour, it could bring the economic benefit and mitigate the emission of waste heat.