設置潛板之定床彎道流場數值模擬研究

Abstract

河川型態中以蜿蜒河川最常見，當水流進入彎道後，形成徑向慣性力以及因水面超高現象之徑向靜水壓差，此兩種作用力不平衡下產生二次流(secondary flow)，進而造成彎道凹岸底床淘刷，凸岸淤積之現象，不僅導致河道變遷，甚至可能危害河川岸邊居民之生命財產。早期大都建造堤防與丁壩保護凹岸，但若該處地質不佳或有大量之底床刷深量，則須付出龐大的工程費，因此設置潛板(submerged vane)之河岸保護工法成為可行替代方案之一，此工法可降低二次流強度、減少凹岸的淘刷與節省興建堤岸的工程費，故本研究探討潛板設置於彎道凹岸後流場分佈。 本研究以許(2002)發展之二維顯式有限解析法模式為基礎，將潛板產生之反二次流效應導入模式中。潛板之主要作用為減少凹岸底床之過度沖刷，因此與動床沈滓輸送現象息息相關，但目前模式是以模擬定床為主，根據de Vriend and Koch(1977)緩彎與Rozovskii(1961)急彎定床之實驗室案例，並假設潛板相關之參數進行模擬。由de Vriend and Koch(1977)模擬案例結果顯示，當每列設置一個潛板時，則最靠近外岸處之流速最大可下降16.3％，而當每列設置兩個與三個潛板時，則最大可下降30％。由Rozovskii(1961)模擬案例結果顯示，當每列設置一個潛板時，流速最大可下降14.6％，而當每列設置兩個與三個潛板時，則最大可下降16％。由此可見，彎道設置潛板後，對凹岸流速分佈有一定程度之影響。此外，模擬Wang(1991)之潛板動床試驗案例，但是以定床進行模擬，以比較動床與定床之流速間差異。由於定床模擬並考慮輸砂現象，兩者間有相當之差距，由模擬結果發現，兩者流速分佈趨勢相似。 由Yeh (1990)之研究中顯示，彎道之二次流強度在發展區會有大於完全發展區之情況，稱之為射出現象(overshooting)，因此為了有效反應二次流強度之變化，根據Yeh (1990)之試值模擬結果，求出二次流流速沿程變化之迴歸式，並採用de Vriend and Koch(1977)與Rozovskii(1961)之實驗室案例進行模擬，發現Rozovskii(1961)案例中考慮射出效應比未考慮射出效應所計算之流速較接近實驗值，而在de Vriend and Koch(1977)案例中，無論有無考慮射出效應，則計算之流速與實驗值幾乎沒有差異。

Sinuous rivers are most commonly seen among the types of natural rivers. When water flows through the channel bends, secondary flow occurs. The secondary flow is driven by the local imbalance between the centrifugal mass force in the radial direction and the hydrostatic pressure generated by the super elevation of the water surface. The effect of secondary flow may cause extra scouring at the outer banks, whereas deposition would occur at the inner banks in the meandering channels. This phenomenon could cause the change of channel, and could even endanger the lives and properties of the residents at the two sides of rivers. In the past, engineers built dykes to protect the outer banks. However, the undesirable geological condition and the presence of massive wash load in the river bed could result in huge engineering cost and thus lower the economic efficiency. The submerged vane is therefore developed in order to reduce the strength of the secondary flow which causes the scouring at the bed in the outer banks and save the engineering building fees of levees. This model is based on the explicitly finite analytic method developed by Hsu in 2002. The effect of the submerged vane is added which can generate the trailing vortex. The primary function of a submerged vane is to reduce the over scouring of the river bed at the outer banks, which is related to the sediment transportation. With assumed parameters of the vanes, this model simulates the fixed-bed experiments carried out by de Vriend and Koch (1977) for 900 bends and Rozovskii (1961) for 1800 bends. By installing a number of submerged vanes in the river bend, the change of velocity at the outer banks is studied. The simulated results in de Vriend and Koch’s (1977) experiments show a maximum drop of velocity of 16.3% upon the installation of one submerged vane per row in the river bed. A drop of velocity of 30% is shown for the installation of two or three submerged vanes per row in the river bed. The simulated results in Rozovskii’s (1961) experiments show a maximum drop of velocity of 14.6% upon the installation of one submerged vane per row in the river bed. A drop of velocity of 16% is shown for the installation of two or three submerged vanes per row in the river bed. Thus there is a substantial influence of the distribution of velocity after the installation of submerged vanes. In addition, Wang’s (1991) submerged vane experiments under the moved-bed condition are simulated using the fixed-bed function in order to compare the differences between the velocities under the two conditions. Despite the considerable differences between the simulations under the two different conditions due to the absence of sediment transportation parameters under the fixed-bed condition, there is a similar trend in the distribution of velocity. Studies from Yeh (1990) discovered the strength of secondary flow is larger in the developing region than in the fully-developed region at channel bends. The phenomenon is called “overshooting”. In order to effectively reveal the strength of the secondary flow, the experiment analogue result according to Yeh (1990) is used and the regression equation of secondary flow is found. Then the model simulates the de Vriend and Koch(1977) and Rozovskii (1961) experiments. It is found that with the consideration of the overshooting effect, the simulated results are closer to the experimental data performed by Rozovskii compared to the ones without the consideration of the effect. For de Vriend and Koch’s (1977) experiments, there are no significant differences between the two considerations.