運用多目標水庫最佳化操作提升水、糧食、能源之協同效益

Using Multi-objective Reservoir Optimization to Improve Synergistic Benefits of Water, Food and Energy Nexus

温庭玄(Tinn-Shuan Uen)

指導教授 ： 張斐章

國立臺灣大學/生物資源暨農學院/生物環境系統工程學研究所/碩士(2017年)

https://doi.org/10.6342/NTU201703371

摘要

近幾十年迄今，都市化對於台灣都市人口的成長以及水、糧食、能源之資源管理影響重大。而自然資源的缺乏在水、糧食、能源這三項自然資源的複雜鏈結關係下，將會是一個長時性、權衡性的艱深問題。由於現今地球環境的變動，綠色能源、綠色環境等相關概念已逐漸成熟，但能源的部分，許多國家仍相當依賴核能發電。在台灣，每年平均有3至5場颱風侵襲，而颱風伴隨著大量的降雨是一種可利用之水資源。在過去的防洪操作中，排洪道與溢洪道所排出的水量相當大，但這些水資源並未充分應用於能源上的使用，造成水資源的浪費。因此，本研究基於提升水、糧食、能源之協同效益，利用台灣先天氣候條件的優勢，提高水力發電之應用，並將水庫操作結合備援供水系統，將缺水率最小化。以下將本研究分為四部分介紹。
第一部分，本研究藉由分析過去20年石門水庫之入流量，探討歷史乾旱年，於不同的初始水庫容量情境下，依據石門水庫操作規線(M-5)操作，結合桃園地區埤塘與石門水庫中庄調整池之備援供水系統，模擬自十月至隔年六月之農業與公共之平均旬缺水率。結果顯示，埤塘與中庄調整池之備援供水系統之平均缺水率改善最高可達25%。
第二部分，本研究透過水庫操作，於颱風時期(七月至九月)最大化水力發電，並同時考慮未來供水風險。本研究設定三個初始水位情境(240、235、230公尺)，選取過去11年包含豐水年、一般年、乾旱年之資料，透過非支配遺傳演算法(NSGA-II)搜尋水力發電量總合及期末庫容總合兩個目標，並將搜尋之最佳化操作結果套用M-5規則進行比較。結果顯示，於豐水年情境下最佳化操作下水力發電量可大幅提升，最高發電收益超過17,000萬元；於M-5限制下，最大發電量仍有15,198萬元之收益。於一般年，最佳化操作之發電最高收益達4,832萬元；於M-5限制下，最大發電效益為2,568萬元，缺水改善率大多維持相同水準。於乾旱年，最佳操作模式則完全根據M-5規線操作，故發電效益相同。
第三部分，本研究將第一部分長期水庫操作模擬，結合第二部分颱風季節之短期操作，探討於最佳化水力發電之後，水庫未來之供水安全。本研究選取過去20年最乾旱之情境2002-2003年，探討水庫於颱風季節操作後之缺水模擬情況。結果顯示，豐水年可改善平均缺水率最高達10.5%；於一般年平均缺水改善率最大達14.2%；乾旱年則與原始操作相同。
第四部分，本研究針對颱風事件之水庫操作進行分析，於颱風侵襲前提前放水發電，設定發電放流量上限為三個情境(68.6 cms、90 cms、120 cms)，並挑選7場洪峰入流量不同之颱風作模擬，利用NSGA-II搜尋發電量與期末庫容，探討不同情境與颱風強度下之發電操作對期末庫容的影響。結果顯示，洪峰入流量高於2000 cms之颱風可全力發電，發電效益最高達4,470萬元；而洪峰流量1000 cms左右之颱風，全力發電時操作期末庫容會降低，最低為12277萬噸。
整體結果顯示，本研究之颱風季節最佳化操作，結合水庫與備援供水系統之長期供水調配，除了能提升颱風季節之水力發電效益，亦可降低乾旱年之農業與公共缺水率，且颱風事件操作分析，未來可強化颱風期間水力發電效益，進一步提升水、糧食、能源資源運用之協同效益。

關鍵字

都市化 ； 水、糧食、能源鏈結 ； 非支配遺傳演算法 ； 最佳化 ； 多目標水庫操作

並列摘要

In recent years, urbanization has significant impact on population growth and resource management of water, food and energy nexus (WFE nexus) in Taiwan. Resource deficit of WFE has become a long-term and thorny issue due to the complex interactions of WFE nexus. As the changes in the global environment, green energy, green environment, and other related concepts have become significant, but many countries still rely heavily on nuclear power. There are averagely 3 to 5 typhoons strike Taiwan every year accompanied by a large amount of rainfall which could be used as available water resource. However, without doing so the extremely large quantity of spilled water for merely flood control purpose has turned to serious waste of water resources. Therefore, in order to improve the synergistic benefits of water, food and energy nexus, we propose to took advantages of climatic conditions in Taiwan to enhance the hydropower efficiency while minimized the water shortage rate of long-term reservoir operation in the future by combining reservoir with water support systems.
In the first part, we analyzed the annual inflows of the Shimen Reservoir in the past 20 years to analyze the most drought years. By setting up different amounts of initial reservoir storage and combining reservoir with water support systems, we simulated the long-term operation from October to next June under the official operation rules of the Shimen Reservoir (M-5 rule), and calculate the average water shortage rate (WSR) of every ten-day for both agricultural and public uses. The results show that the average WSR can be reduced up to 25%.
In the second part, we maximized hydropower generation during typhoon season (July to September) while considering risks of future water supply through reservoir operations. We set three scenarios for initial water level (240, 235, 230 meters) and selected data from the past 11 years, including wet, general, and drought year, and applied Non-Dominated Sorting Genetic Algorithm (NSGA-II) to search for the total amount of hydropower generation and the final reservoir storage. In addition, we compared the optimized operations with those constrained by the M-5 Rule. The results show that the optimized operations enhance hydropower generation greatly in wet year, the highest benefit from the hydropower is more than NT$ 170 million by optimized operation and still NT$ 150 million under the M-5 Rule. In general years, the highest benefit is NT$ 48.32 million with optimized operation and NT$ 25.68 million under the M-5 limit. In drought years, the optimized operation is same as M-5 rule operations.
In the third part, we integrated the long-term reservoir operation simulation in the first part and the short-term operation during typhoon season in the second part into a full year operation, analyzing future water security based on the optimal hydropower operation. To explore the water shortage simulation after typhoon season operation, we selected the driest situation in the past 20 years. The results show that the average WSR in every ten-day can be improved up to 10.5% in a wet year and up to 14.2% in general year. However, the average WSR in a drought year is the same as the original operation.
In the fourth part, we focused on the reservoir operation based on typhoon event, in order to maximize the hydropower generation before a typhoon strikes. We set up three scenarios for the upper limit of water discharge rate for hydropower (68.6, 90, and 120 cms), and selected 7 typhoon events for different peak inflows, searching for total hydropower generation and the final reservoir storage by NSGA-II with the hourly-based time-scale. The results show that it’s feasible to operate for full capacity hydropower and final reservoir storage when the peak inflow is higher than 2000 cms. The benefit for hydropower is up to NT$ 44.7 million. However, when the peak flow is about 1000 cms, the final reservoir storage decreases, and the minimum is 12.277 million tons.
In conclusion, the operation strategy that combined the optimized operations during typhoon season with the long-term operation can not only enhance the effectiveness of hydropower but also reduce the WSR in a drought year. Furthermore, analysis of typhoon event operation strengthened the benefits of hydropower during one single typhoon event, and eventually ameliorate synergistic benefits of resources allocation among water, food, and energy nexus.

並列關鍵字

Urbanization ； Water, food, and energy nexus ； Non-Dominated Sorting Genetic Algorithm (NSGA-II) ； Optimized ； Multi-objective reservoir operation

參考文獻

28. 蔡昀直，2012，供水系統綜合可靠度分析-以石門水庫為例，國立交通大學土木工程學研究所學位論文，1-95。

21. 王國威，2006，運用懲罰機制遺傳演算法於水庫颱洪操作之規劃，私立淡江大學水資源及環境工程學研究所碩士論文，1-76。

27. 黃仁政，2004，水庫操作規線優選模式建立與求解，國立臺灣大學土木工程學研究所學位論文，1-80。

24. 楊舜年，2015，建立颱洪時期抽水站智慧型最佳化操作規則，國立臺灣大學生物環境系統工程學研究所學位論文，1-126。

22. 王昱中，2014，智慧型水資源調配策略以因應用水需求成長，國立臺灣大學生物環境系統工程學研究所學位論文，1-92。