本論文希望藉由數值模擬來率定不同移動頻率和不同阻塞比水平向TLCD受水平力移動後液面振幅高度，將模擬之結果與前人之實驗結果做比較，以找出最佳預測水平向TLCD液面振幅高度之數值模式。以往率定TLCD之方法都是以實驗為主，實驗模型及量測儀器都是必要的花費，若能設計出高準確性之數值模式取代實驗，就能大幅降低研究成本，也能節省人力及研究時間。 本研究使用計算流體力學軟體Fluent中滑動網格與VOF兩種技術的結合，有效地預測TLCD受水平力移動後液面擺動之振幅，並計算出不同移動頻率和不同阻塞比之水頭損失係數及找出TLCD之自然頻率。紊流模式採用層流模式及 標準紊流模式進行測始，結果發現使用層流模式預測最大液面振幅較 標準紊流模式準確。而在模擬中發現左右兩個垂直柱分別之最大液面振幅有明顯誤差，將網格局部加密後發現誤差值變小，且60~80秒內之平均值不會因此而改變，因此為了增加模擬效率，採用均一結構網格大小，而不採用局部網格加密。 由數值模擬結果與實驗結果比較得知，靠近共振頻率時之最大液面振幅誤差較大，誤差百分比約為10%，其餘移動頻率誤差百分比則為5%以下,而由數值模擬求得之水頭損失係數與TLCD自然頻率之結果和實驗值比較誤差不大，在可接受之容許範圍內。 The purpose of this thesis is to find a better CFD model, the results are compared to experimental measurements, to predict the amplitude of water in the different horizontal movement frequency and oriffice TLCD. Previous studies of TLCD were done by experiments, and it that spend much more money for experimental model and apparatus. If a CFD Model with high accuracy is found to replace the experiment, not only saving the experimental cost, but also the time and the manpower.
Functions of sliding mesh and Volume of Fluid are included CFD models. They are efficient to predict the amplitude which is the result from horizontal force in the TLCD and to calculate the head loss coefficient and the nature frequency of different movement frequency and orifice TLCD. The standard model and the laminar model were adapted to beginning the test. The result of the test was found that the laminar model is more accurate for predicting the amplitude of the largest water surface. In this study, a problem was found that the largest amplitude was not the same at two side vertical columns, and therefore the cell near the water surface was going to adapt for detail. Afterward the amplitude for two side column will be closed, and the average of the amplitude would not be changed at sixty to eighty seconds. To allow for efficiency the cell will adapt for original.
By the comparison between CFD and experiment results, the error is larger about ten percent near the resonance frequency case and five percent for others. However, the head loss coefficient and the natural frequency which is solved by CFD are closed to experiment, and the error is acceptable.
