Abstract
國內具有多機組的汽電共生工廠,目前皆未進行精確的經濟調度分析,本 論文試針對中油林園廠進行此項分析,中油林園廠目前共裝置四座發電用 高壓蒸汽鍋爐,與四部單機發電容量為10 MW的背壓式汽輪發電機組,因年 久運轉,各鍋爐與各發電機組的熱耗值,不同於原廠提供的曲線數據,因此, 本研究於民國84年利用廠內三輕工場停爐歲修,執行蒸汽鍋爐的降載運轉 與發電機的昇降載,以蒐集其完整的熱耗值資料,再運用類神經網路法擬合 各曲線,據以計算擬合函數的係數,用之進行鍋爐與發電機組的經濟調度, 前者以滿前足全廠蒸汽總需量,抑低鍋爐總耗燃熱值為目的,後者在耗用固 定的蒸汽熱量條件下,以提高全廠總發電量為目的,經數值驗證,方法可行, 具有經濟效益,根據測試結果,對於現行調度運轉方式提出建議,民國 85年 底林園廠將擴增完成一座容量為40MW的抽汽/冷凝式汽輪發電機組,與蒸汽 容量為250噸/小時的高壓蒸汽鍋爐,屆時調度更有彈性,預期調度效益將更 顯著. This thesis presents a modeling approach to the economic dispatch of multi-unit cogeneration system of Lin-Yuan plant of Chinese Petroleum Corporation (CPC). The model to be presented is the first trial model of the cogeneration system in Taiwan. The cogeneration system of Lin-Yuan plant comprises of four high-pressure steam boilers and four back-pressure steam turbine generators each of 10 MW. Because these boilers and turbine generators have been operated for years, their heat rate curve already differed from their original manufacturers' design data. To measure the heat rate, we carried out a series of tests by uploading or deloading these units during the annual maintenance period of 1995 for the third naphtha cracking plant of Lin-Yuan. We then curve-fitted the heat rate measurement data by the artificial neural network method. Based on the heat rate function obtained by these curve-fitting, we constructed the economic dispatch model for the boilers and turbine-generators respectiv- ely. The boiler model aims to decide the steam production of each boiler at the least total fuel consumption to meet the steam demand of Lin-Yuan plant. The turbine-generator model decides the power production of each turbine-generator to maximize total power production under a fixed amount of steam supply. According to the numerical testing results of these two models, we evalua- ted the present operation of Lin-Yuan plant and proposed several economic operation suggestions. By the end of 1996, CPC will install, at Lin-Yuan, one more high-pressure steam boiler with size of 250 tons/hr and one turbine generator of the extraction/ conde- nsation type with size of 40 MW. With this additional flexibility to the plant operation, our modeling approach will find itself a more significant impact on the cogeneration operation.