南化水庫最佳化與模糊化營運規線對缺水影響探討

Impacts of Optimal and Fuzzified Rule Curves on Water Deficits for Nanhua Reservoir

洪彥甯(Yen-Ning Hung) ； 蕭政宗(Jenq-Tzong Shiau)

《農業工程學報》 63卷2期 (2017/06) Pp. 67-77

https://doi.org/10.29974/JTAE.201706_63(2).0006

摘要

台灣南部地區由於降雨時間分布不均且地勢陡峭、河川湍急，故須水庫調節農、工業及民生用水需求，如何有效運用水庫供水以滿足各方需求是水資源管理的一項重要議題。目前水庫操作多依循營運規線(rule curve)，以評估不同時期水庫蓄水量之多寡，當蓄水量低時，會以減少供水的方式降低未來發生嚴重缺水的機會。然而，現有水庫營運規線因個數較少而產生階梯式之限水方式、或是早期制定的規線無法掌握目前極端的水文狀況。為改善現有水庫營運規線所造成的缺水現象，本研究考慮二優選模式，模式一為模糊化營運規線，模式二為最佳化營運規線。本文以南化水庫為例，以多準則決策方式整合8 個缺水指標(單旬最大缺水率、缺水事件最長缺水延時、最大缺水量、平均缺水延時、平均缺水量、風險度、總水率、缺水事件頻率)為優選模式目標函數，配合遺傳演算法尋求最佳解，並與現況營運比較。分析結果顯示，二優選模式之供水效率均優於現行規線，而其中又以最佳化規線之結果較優，在8 項缺水指標中有6 項優於現況、7 項優於模式一。

關鍵字

水庫規線 ； 優選模式 ； 模糊理論 ； 多準則決策

並列摘要

Southern Taiwan had suffered from water deficits due to uneven temporal and spatial distribution of rainfall. Reservoirs are important facilities to regulate fluctuating streamflow to offer stable water supplies in this region. Water releases from reservoirs are often guided by rule curves in Taiwan for assessing storage in various seasons. Water rationing is implemented to conserve water for future use in order to reduce possibilities of high deficits when water below certain level. However, finite rule curves leading to step-wise hedging rules as well as pre-defined rules curves insufficient to meet current hydrologic conditions are major reasons for low water-supply efficiency. To improve reservoir performance during droughts, two optimization models are considered in this study to construct optimal operating rules for Nanhua Reservoir located in southern Taiwan. Model I fuzzifies the current rules curves and Model II reconstructs the lower rule curve of the Nanhua Reservoir. A total of 8 water-shortage indices (maximum 10-day shortage ratio, maximum event duration, maximum event deficit, average event duration, average event deficit, risk, total shortage ratio, event frequency) are integrated into a single overall objective to derive the optimal operating rule using multi-criteria decision making. The results indicate that both optimization models have better performance than the current operation rule and Model II outperforms Model I.

並列關鍵字

Rule curves ； Optimization operation ； Fuzzy theory ； Multiple criteria decision making