運用基因演算法探討低衝擊開發之空間配置策略─以台大校園為例

隨著都市不斷的開發，地表逕流與洪峰流量隨著不透水的面積急遽增加，以往集中式的排水下水道已漸漸無法應付氣候變遷所帶來的強烈降雨，採用分散式就源處理的各種概念為可能的解決方式，其中又以低衝擊開發設施（LID）最廣為人知，其概念在於增加都市的綠地以回復至開發前的保水入滲效果。過去已經有不少的研究指出生態滯留單元與透水鋪面等LID，在台灣等亞熱帶雨型之下是較為有效，但過去的研究僅驗證在大量設置LID的情況下有一定的效果。然而在預算、空間有所限制的情況下，不能再如同以往未經規劃盲目設置LID，如此將有可能付出金錢、時間、人力的成本，卻無法獲得LID所帶來的減洪效果。為此，我們必須了解LID削減洪峰之機制為何，唯有掌握其極限所在，方能做出正確的容量、負擔設計，此外，為了解在進行社區、都市尺度的設計時，LID應該優先設置何種區域，必須先分析LID集水區空間配置之特性。　　本研究以臺大校總區為研究區域，以36種降雨情境，透過美國環保署所開發的暴雨逕流管理模式（SWMM）及其LID元件進行模擬。從單一次集水區的敏感度分析去探討LID設施削減洪峰之機制為何。再調整生態滯留單元以及透水鋪面於各次集水區之配置面積與負擔比例，透過基因遺傳演算法（GA）結合SWMM進行最佳化，最後以單位歷線對次集水區進行水文分析，以了解如何做出較佳之LID空間配置。研究結果發現在運用SWMM模擬計算時，由於模型的簡化假設必須要留意其LID元件及其溢淹機制與實際情況有所出入。而LID之削減洪峰效果視其在洪峰來臨之前是否仍有足夠蓄水空間而決定，在設有土壤層的生態滯留單元，於台灣較為強烈的降雨條件下以表面的蓄水空間為主，沒有設置土壤層的透水鋪面其蓄水層可以發揮很好的減峰效果，但在較大雨勢容易蓄滿，依然是以地表蓄水空間為主。而最佳化的結果顯示，LID應優先配置在主流中游，其次為主流下游，不建議配置於主流或支流上游。透過單位歷線一系列的水文分析，發現配置較多LID之次集水區與其削減洪峰機制相吻合。

關鍵字

低衝擊開發；最佳化；空間配置； SWMM ；基因遺傳演算法

並列摘要

With growing of impervious areas due to urban development, surface runoff and peak flow become larger and larger gradually. Under the impact of climate change, the current drainage system seems have more difficult to deal with strong and short duration rainfall which cause serious threat to urban area. As a result, distributed and source control of runoff strategy are proposed as a powerful solution. Low impact development （LID） is well known as one of source control. Its concept is using “functionally equivalent hydrologic landscape” to improve ability of infiltration and retention to make condition of urban surface return to pre-development. In practical design, engineers always need to set up LID with effective plan under constraints of budget and space. Thus, we must equip the knowledge of the mechanism in peak flow reducing by LID. Besides, we should also analyze the spatial arrangement of LID and understand the most effective layout of LID in a watershed. 　　This study took main campus of National Taiwan University （NTU） as objective area and simulated 36 kinds of rainfall by US EPA SWMM5.1. Through the sensitivity analysis of setting LID in single catchment we could find the mechanism of reducing peak flow by LID. After setting area and load of LID as variables uses genetic algorithm （GA） combined SWMM and analysis of unit-hydrograph to find out the best spatial arrangement of LID. Through results find that the simulation method of LID elements and flooding in SWMM model have some difference from reality. It is worth to note that the rest of capacity in LID before peak flow determines whether LID reduces peak flow or not. Capacity in both bio-retention and permeable pavement are not dominated by storage layer. Because bio-retention has soil layer which control speed of percolation makes water deep into storage layer slowly, capacity is dominated by surface berm height. Although permeable pavements do not have soil layer, capacity are still dominated by surface berm height due to its easily filled storage layer. Besides, the results from optimization shows that LID should set at midstream first and downstream second, and setting LID at upstream seems workless. Finally, through analysis of unit-hydrograph found that the spatial arrangement of LID corresponds to its mechanism indeed.

並列關鍵字

Low Impact Development ； Optimization ； Spatial Arrangement ； SWMM ； Genetic Algorithm

參考文獻

[5] 何媚華. (2014). 中永和地區都市排洪系統最佳管理措施之探討. 臺灣大學土木工程學研究所學位論文, 1-95.

[7] 栗晨堃. (2016). 都市雨水貯留設施設計優化模式發展與應用—以臺北市民生社區為例. 臺灣大學土木工程學研究所學位論文, 1-98.

[8] 黃耀賢. (2015). 都市低衝擊開發設施最佳化配置研究─ 以臺北市民生社區為例. 臺灣大學土木工程學研究所學位論文, 1-105.

[6] 徐硯庭. (2014). 低衝擊開發運用在高都市化地區的減洪效益-以新北市中永和地區為例. 臺灣大學土木工程學研究所學位論文, 1-126.

[10] 褚筱文. (2010). 應用 BMPDSS 模擬都市暴雨之最佳化管理.

國際替代計量

運用基因演算法探討低衝擊開發之空間配置策略─以台大校園為例

全文下載

主題瀏覽