Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/14108
標題: 混合礫石陡坡渠槽之非平衡篩選現象研究
Nonequilibrium Sorting of Nonuniform Gravel in a Steep Channel
作者: 蘇志強
Su, Chih-Chiang
關鍵字: steep channel;陡坡渠槽;non-uniform gravel;non-equilibrium sediment transport;grain-sorting;artificial neural network;混合礫石;非平衡輸沙;顆粒篩選;類神經網路
出版社: 土木工程學系
摘要: 
山區河川及台灣短峻湍流富饒粒徑寬廣之卵礫石,洪水時顆粒篩選現象時常發生。九二一大地震後山地蘊藏大量土石,且全球天候暖化後,颱風豪雨趨於超強度集中降雨,如2004年七二大水,導致台灣中部大甲溪河道過載淤積,影響國計民生。
本文利用碩士論文平衡輸沙資料,進行非平衡沖淤之床沙分離現象探討。以相同粒徑範圍、泥沙中徑而幾何標準偏差為2及1.5之兩組混合礫石;變化兩組坡度範圍為2%-4%與3%-5%;試驗過程採固定流量及加沙率,設計泥沙供給量為過載、弱載,共計8個非平衡沖淤全程試驗,各全程試驗另含4個分段試驗,總計完成40個陡坡非平衡試驗。
根據全程試驗結果,本文認為水流篩選機制,包括沿主流向重力篩選及垂直向隱蔽篩選。泥沙幾何標準偏差愈大且底床坡降愈陡之非平衡沖淤試驗,重力效應導致粗顆粒移動性逆轉和床沙細化影響水流強度等愈為顯著。過載淤積試驗,推移載及過載淤積層之泥沙中徑,沿主流向下游粗化;隨底床淤積而底床坡降趨陡,受重力效應影響之粒徑趨小,淤積層粒徑垂直向上趨於分層細化。
由分段試驗結果,本文驗證非平衡試驗具可重複性;且分段試驗之床沙糙率分析,可驗證水流重力及隱蔽篩選現象。此外,本文收集礫石之渠槽平衡試驗資料,利用類神經網路,建構陡坡礫石推移載之倒傳遞神經網路模式,已獲致良好驗證。

Mountain rivers such as those in the upland areas of Taiwan usually have gravel with wide gradations. The sorting phenomenon frequently occurs in the bed material during the floods.. After the ghastly 921 earthquake, the mountainous region contains a large quantity of loose sediment. In addition, with the advent of global warming, the torrential rain carried by typhoons tends to fall intensively within a specific area. For example, the heavy rainfall on July 2, 2004 caused the Da-Chia River in the central Taiwan to overload with gravel, and subsequently affected the national economy and the people's livelihood enormously.
In this study, the selective transport phenomenon of the non-uniform gravel under the non-equilibrium condition was investigated according to the equilibrium condition data from the author's Master's thesis. With the geometric standard deviations of 2 and 1.5, the same size ranging from 2.36 to 38.1 mm, and the same median size of 7.5 mm, two kinds of particle size distributions were prepared by sieve analysis of the natural gravel. As a result, forty non-equilibrium tests (including the total runs and the subruns) were conducted under both the underloading and overloading conditions with a slope ranging from 2% to 5%.
According to the experimental results, the grain-sorting mechanism includes both the gravity sorting along the longitudinal direction and the vertical sorting by the hiding phenomenon. As the non-equilibrium experiments were carried out under the conditions of larger geometric standard deviation and steeper bed slope, phenomena such as the mobility reversal due to the gravity effect, and the flow intensity influenced by the fining of bed material, became more obvious. The downstream coarsening of the median size in both the bed load and the deposited layer occurred mainly due to gravity effect in aggrading laboratory deposits. In addition, the lower limit of the coarse grains affected by the gravity effect decreased with an increase of bed slope (mobility reversal), so the upward fining of the deposited material also occurred due to the hiding phenomenon.
The sorting phenomenon was verified by the analysis of the roughness of the aggrading deposits in the subruns. In addition, developed from the ANN (artificial neural network) model for predicting the bed load with non-uniform gravel in steep slopes in this study, the BPN (back-propagation network) scheme was also proved to be reasonably accurate with flume data collected under the equilibrium condition.
URI: http://hdl.handle.net/11455/14108
Appears in Collections:土木工程學系所

Show full item record
 

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.