在高壓區域內螺旋凹槽式真空邦浦之流場計算

Abstract

本文是以計算流體力學的方法在ALE(arbitrary Lagrangian-Eulerian)座標系統下分析螺旋凹槽式真空邦浦內動態網格產生的流場。計算時使用非結構性和非交錯式網格；對於速度與壓力的偶合則採取SIMPLE法則並以有限體積法將統御方程式離散化。而為了消除非交錯網格產生的棋盤式震盪(checkerboard)則使用了Rhie & Chow的方法修正面上質量流率。而由於模擬此流場時是固定進出口壓力以獲得邦浦之流量，所以需對入口的壓力邊界條件作特別的處理。 本文針對T. Sawada【9】的螺旋凹槽式真空邦浦進行分析，主要是探討在層流狀態下此一幫浦的性能。在研究中測試了不同的設計參數對於抽氣效能的影響。結果顯示流道角度與流道高度的改變對流量會有正面及負面的效應，因此在角度與高度間均存在一最佳化的值。間隙的存在會有餘隙洩漏的效應，所以當間隙愈大時邦浦的性能將愈差。當流道數目愈少時其受間隙的洩漏影響減小，所以流量隨著流道數目減少而增加。當轉速加大時能使得壁面傳遞更多的動量至流體因此產生了較大流量。 另外本文以線性κ–ε模式配合壁函數來分析此邦浦的紊流流場，而由於使用了高雷諾數的紊流模式因此與實驗的結果有所差異。為了克服此一問題可能須採取適用於低雷諾數的紊流模式，然而這也增加了數值模擬的複雜度，所以此類問題仍有待解決。

In this thesis, CFD is used to analyze viscous flow of dynamic meshes in spiral grooved vacuum pumps with ALE method. Unstructured and nonstaggered meshes are adopted in computations. The SIMPLE algorithm is taken to couple velocity-pressure relations and the governing equations are discretized by finite volume method. For avoiding checkerboard from nonstaggered meshes, we correct mass flow with a Rhie & Chow method. Besides, a throughput of the pump is obtained by fixing the inlet and outlet pressure, and a special treatment of velocity-pressure couple is necessary for the pressure boundary in the inlet. In this thesis, a spiral grooved vacuum pump [9] is analyzed and we explore the pumping performance in the laminar flow mainly. Different design parameters are tested how to influence the pump throughput. Testing on spiral angle and channel height indicates that these parameters need to be optimized to achieve better performance. The appearance of clearance leads the gap leakage and the pumping performance is becoming worse with increase of the clearance. When number of channel is fewer, influence of the gap leakage is lowered and the throughput increases with reduction of number of channel. Results also reveal that increase of rotating speed transfers more momentum into fluid and the larger throughput is obtained. As regards turbulent flow computations, the linear κ–ε model and wall-functon are incorporated to solve. The difference of computation and experiment is possibly due to the use of high Reynolds turbulent model. To overcome this problem, a low Reynolds one is adopted to solve but the complexity of numerical model is also increased. This problem is still needed to solve.