考量減洪效果下中和地區最佳雨水貯集系統設計

Optimal Spatial and Volume Design of Rainwater Catchment Systems for Flood Mitigation in Zhonghe Dist.

羅偉峻(WEI-JIUN LUO)

指導教授 ： 徐年盛

國立臺灣大學/工學院/土木工程學研究所/碩士(2013年)

https://doi.org/10.6342/NTU.2013.01233

摘要

目前世界在極端氣候下降雨更加不平均，遇雨則澇、缺雨則旱。低衝擊開發在美國推廣已約12年，應用上較廣，所採用的低衝擊開發技術包刮：雨水貯留筒、藍屋頂、綠屋頂、透水鋪面、道路旁之生態滯留空間及雨水花園等。為了降低劇烈暴雨帶來的洪峰及總逕流量，低衝擊開發設施與傳統防洪措施建造地下管線涵洞來比較，低衝擊開發成本相對較低，且可帶來其他附加價值，如涵養水源、再生水利用、美化城市、改善生態環境等。在台灣之都市排水目前還是以雨水下水道為主，一般設計標準為2~5年重現期之暴雨，低衝擊開發技術在台灣還沒有真正形成，目前僅有個別之低衝擊開發技術，如雨水貯留桶、綠色屋頂等，但減少暴雨逕流和延遲峰值的效果尚未完整評估。
本研究由暴雨逕流管理系統(Storm Water Management Model，簡稱SWMM)中導入低衝擊開發(LID-Low Impact Development)技術中之雨水貯留系統，並優先選定近年水災頻繁的中和地區，將低衝擊開發策略之理念應用於中和。比較目前實際情況與雨水桶設置後之爆管量減少情形，並由雨水桶的大小及密度分布評估計算其淨效益，期望找出最適合此中和地區最佳的雨筒尺寸與雨筒設置位置。
本研究建立一套模擬-優選模式，以快速自動化地優選出淹水損失並最大化淨效益下都市排水區域之雨量筒空間布置與容量之最佳設計方式。於模擬法中，以實際降雨資料輸入建構完成之SWMM模式以進行模擬排水管水位與爆管量，並於SWMM中設計不同的雨水貯留設施空間分布及容量，進而得知各種設計情況下減洪情況，並進一步計算其扣除成本後之淨效益。優選法中，首先由模擬法所得之資料輸入倒傳遞類神經網路(BPNN)進行訓練與驗證，得到一可代替SWMM模式之BPNN水位模擬模式，並將BPNN模式嵌入優選模式，並與禁忌演算法結合以優選出最佳化淨效益之雨水貯留設施空間分布及容量。
結果顯示應用禁忌演算法求解優選模式最佳解可比模擬法更佳12.75%，故優選法優於模擬法。而由於SWMM模式屬於視窗化之商業軟體，難與優選模式與優選方法結合。然而本研究應用所研發代替SWMM之BPNN模式嵌入優選模式中，結果顯示可快速且有效地搜尋到最佳解，因此顯示本研究所建立之方法可解決前人研究之限制。(3) 於模擬法中，所求之最佳解年淨效益為13,841,832元，其各分區雨筒數量分別為0、50、236、109，尺寸均為12立方公尺。於優選法中，所求之最佳解年淨效益為18,302,374元，各分區雨筒數量分別為0、30、206、99，尺寸分別為0、13、11、8。並將優選法之空間分布與容量設計代回SWMM模式，經過此雨筒設計可減少72%淹水損失，減去年成本後得到之年淨效益為15607090元。其結果也顯示本研究所研發模擬-優選流程可優選到較具彈性與實用性之雨水貯留設施空間布置與容量設計方式。

關鍵字

低衝擊開發 ； SWMM ； BPNN ； 禁忌演算法

並列摘要

The rainfall become more uneven in extreme climates in recent years, It's either a drought or a flood. Low-impact development in the United States had promoted for 12 years, and with wider application. Low-impact development including: rain barrels, blue roofs, green roofs, permeable pavement, and bioretention on roadside space. In order to reduce violent storm brought peak and total runoff, compare with traditional flood control measures in underground pipeline construction culvert, low-impact development costs relatively low, and can bring additional added value, such as water conservation, recycled water use, beautify the city, improve the ecological environment. The urban drainage to storm sewer main is general designed criteria of 2 to 5 year return period storm currently. Low-impact development techniques in Taiwan has not really formed, only the individual low-impact development techniques at present, such as rain barrels, green roofs, etc., but to reduce the peak stormwater runoff and delayed effects is incomplete assessment.
This study was funded by stormwater runoff management model(referred SWMM) to import low-impact development(LID) technology of rainwater systems, and prioritized in Zhonghe Dist., which frequent floods in recent years, then the concept of low-impact development strategies used in Zhonghe Dist. Compare the current situation with the rain barrel set up after storm pipe reduced circumstances, by the size and density of the rain barrel distribution, calculate the net benefit assessment, expect to find
the best rain barrel rain barrel set size and position which was the most suitable in Zhonghe Dist.
Therefore, expect to establish a set of simulation - optimization model. The simulation method, use the actual rainfall data input built SWMM model simulations, and in SWMM in designing different spatial distribution of water storage facilities and capacity, and thus a variety of design situations that reduce flood situation, and further calculate the net benefit after deducting costs. On optimizing method, the information obtained by the simulation input neural network which is training and validation to give a neural network alternative to SWMM model. And optimizing model embedded neural network model, tabu algorithm to find the optimal solution by optimizing method, and compare optimization and simulation method pros and cons.
The results showed that application of tabu algorithm can find better solutions than simulation method, It is optimizing method is superior to the simulation method. SWMM model because the software is windows based, it is difficult with the optimizing mode of binding, in a optimizing mode of BPNN embed this method resolves the limitations of previous studies. The simulation method, the optimal solution are asking for the NT13,841,832 annual net benefits, the number of rain barrel of the district were 0,50,236,109, dimensions are 12 cubic meters. On optimizing method, the required annual net benefits of the optimal solution is 18302374, the number of rain barrel of the district were 0,30,206,99, dimensions are 0,13,11,8 cubic meters. And the optimizing method for the spatial distribution and capacity design substituting SWMM model, after this rain barrel design reduces 72 percent flood damage, annual net benefits is NT15607090 which reduce annual cost.

並列關鍵字

LID ； SWMM ； BPNN ； TABU Search.

參考文獻

22. 黃建霖，「智慧型都市排水抽水站及時操作系統之研發」，農業工程學報，2012。

23. 褚筱文，應用BMPDSS模擬都市暴雨之最佳化管理，國立台北科技大學土木與防災研究所碩士學位論文，2010。

1. Abi Aad, Maya P., Suidan, Makram T., Shuster, William D. “Modeling Techniques of Best Management Practices: Rain Barrels and Rain Gardens Using EPA SWMM5,” Journal of Hydrologic Engineering, P.434-443, 2010.

2. Ackerman D, Stein E D, ”Evaluating the effectiveness of best management practices using dynamic modeling,” Journal of Environmental Engineering, P.628-639, 2008.

3. Ahiablame, Laurent M., Engel, Bernard A. “Effectiveness of Low Impact Development Practices: Literature Review and Suggestions for Future Research”, Water Air Soil Pollut, P.4253-4273, 2012.