懸浮泥砂之對流、擴散及隨機運動機制之探討

Abstract

泥砂運輸與人類的生活息息相關，例如橋墩沖刷、水質估計等等，故泥砂運輸的研究是一直以來都是一個很重要的議題。泥砂顆粒在水中除了隨著水流方向運動之外，也會因為受到紊流的影響而向周圍不規則擴散，此外，其運動的行為可以馬可夫鏈(Markov chain)來近似，因此本研究將泥砂顆粒的運動視為一個隨機過程。本文以力學原理結合序率方法(Stochastic method)來模擬泥砂顆粒在水中的運動軌跡，亦即增強隨機微分方程中的物理性質，使之更貼近自然情形。 為模擬泥砂顆粒的運動行為，本文以朗之萬方程(Langevin equation)為原型所推導出的隨機擴散粒子追蹤模型(Stochastic Diffusion Particle Tracking Model)呈現顆粒運動因紊流而造成的不確定性。其中，隨機擴散粒子追蹤模型主要包含兩種基本元素：平均漂移項(Mean drift term)，即為顆粒隨著水流方向運動；紊流項(Turbulence term)，即顆粒受到紊流作用而有不規則的運動，也稱為布朗運動(Brownian motion)，係利用維納過程(Wiener process)來模擬。 本研究利用隨機擴散粒子追蹤模型來模擬在一般流況下泥砂顆粒的運動軌跡， 分別使用兩種隨機擴散粒子追蹤模型: 單顆粒粒子追蹤模型(One-particle Particle Tracking Model)和雙顆粒粒子追蹤模型(Two-particle Particle Tracking Model)去模擬。其中，雙顆粒粒子追蹤模型比單顆粒粒子追蹤模型多考慮了顆粒在距離相近時候的變化，因為大尺度的渦流(Large scale turbulence)的關係可能使彼此相近的顆粒具有相似的隨機運動。另外，以巨觀的角度去觀察顆粒整體的運動，可以計算出水中的泥砂濃度，且因為泥砂顆粒受到紊流擾動的影響，使得泥砂的濃度也具有不確定的變化。因此本文呈現顆粒軌跡和泥砂濃度的平均值和標準差來表示泥砂顆粒在水中的不確定性。本研究首先和實驗資料比對單顆粒和雙顆粒粒子模型所估計的濃度以驗證模型的可行性，最後使用此模型分別探討層流流場中和紊流流場中顆粒隨機運動的情形，結果顯示在紊流流場中顆粒的隨機運動比較明顯，因此在高雷諾數(Reynolds number)的流場中估計泥砂濃度時，應考慮漩渦對泥砂顆粒所造成的隨機變化，並給予濃度變動範圍較為恰當。此外，泥砂顆粒運動具有馬可夫特性也在本文中證實。然而，如本文結果所顯示，泥沙顆粒的移動距離卻不是和時間呈線性的正比關係，並不符合菲克擴散(Fickian diffusion)。泥砂顆粒具有再懸浮的現象可能導致泥砂擴散為反常擴散(non Fickian diffusion or anomalous diffusion)。

Sediment transport is an important issue for human. It is closely related to human society, such as bridge scour and water quality. A sediment particle in flow not only follows the flow direction, but also diffuses through the surrounding water due to turbulence. Markov chain is used to approach the movement of sediment particles. From this perspective, particle movement is regarded as a stochastic process in our study; moreover, the proposed models simulate particle trajectories based on stochastic methodologies and physical mechanisms, underscoring mechanics in the stochastic differential equation. To simulate sediment particle movement, the stochastic diffusion particle tracking model (SD-PTM) has been derived from the Langevin equation, which is able to show the random characteristics of sediment movement. SD-PTM has two basic elements, the mean drift term and the turbulence term. One of the particle characteristics, the mean drift term, is that particles follow the flow direction; another one is called the turbulence term that describes random behaviors caused by turbulence diffusion. This movement is known as Brownian motion. In general, the diffusion movement is modeled by the Wiener process. The aim of this study is to simulate sediment particle trajectories under the normal flow condition by the SD-PTMs, one-particle PTM and two-particle PTM. The difference between the single particle model and the paired particle model is that the paired particle model accounts for large eddy turbulence. In other words, the paired particles may have similar random movement if the locations of particles are in the immediate vicinity of each other. Besides, to observe assemblage of particles’ motion in the macroscopic manner, the sediment concentrations can be estimated. Moreover, sediment concentrations involve the property of uncertainty on account of sediment particles’ stochastic trajectories. Therefore, to demonstrate such uncertainty of sediment particles, the ensemble means and ensemble standard deviations of sediment trajectory as well as concentrations are presented in the study respectively. The proposed models are validated against experimental data by ensemble mean velocity and sediment concentrations. Moreover, this study also discussed the random movement of sediment particles under various flow conditions, laminar cavity flow and fully developed turbulent open channel flow. Results show that the random movement of sediment particles is significant in turbulent flow. Thus, it is appropriate to consider the fluctuation of sediment concentrations under high Reynolds number flow conditions. Besides, the Markovian property of the PTMs is validated in our study. However, the variance of particle displacement and time are not a linear proportion as the result. Resuspension of sediment particles may cause particle movement to be anomalous diffusion.