程設計之最佳化過程 動化—以平衡流量計設計為例

Abstract

混合推進系統設計近年來引起大家的高度興趣，混合火箭推系統所需之液態燃料使用 流量計來控制，平衡流量計是所有流量計中最吸引人的，因為其所需之反應時間短且安裝空 間小，在眾多形式之平衡流量計中，孔口板又因重量輕、價格便宜、安裝容易而被廣泛使用， 然而，孔口板的幾何外型使流體經過時瞬間縮小及擴張的現象造成壓力損失比其他平衡流量 計多，此篇論文以基因演算法作為最佳化方法，設計一最佳化平台可設計出於試驗中具有足 夠壓差及較少壓力損失之孔口板平衡流量計。 本論文探討最佳化流程並使用開源 CFD (Computational Fluid Dynamics ) OpenFOAM C++ 程式以平衡流量計為例談討並實作最佳化設計，OpenFOAM 開源程式碼可以解決多種 流場問題，例如可壓縮流、不可壓縮流、兩相流等等，本論文所探討之平衡流量計為不可壓 縮流且雷諾數在 Re=10,000，在高雷諾數下流體在管流中為紊流，本篇論文採用 SST k-ω 紊 流模式求解紊流流場，紊流模式在本篇論文中以流體流過一平做為驗證。本論文以壓差 0.06bar 及 0.09bar 為孔口板平衡流量計之最佳化目標，基因演算法使用 SUS (Stochastic Univseral Sampling)方法做挑選及使用 BLX-α 的交配方法在 10 個迭代內已可以獲得非常 接近欲最佳化之目標。 本篇論文最佳化之目標為設計出特定壓力差之平衡流量計之幾何外型，本論文主要分 為四大部分:分別為統御方程式介紹、 OpenFOAM 流場演算法驗證、最佳化設計方法使用 基因算算法及平衡流量計最佳化設計之結果與討論。

Hybrid propulsion has attracted tremendous attention in recent years. The flow rate of hybrid propulsion system is measured by on-board flow meter, in which balanced flow meter is one of most attractive choice because of its very short response time and very limited space required. Among these, the orifice plate features light weight, cheap price and easy installation which make it become the most popular one among all type of balanced flow meters. However, the pressure loss caused by the traditional orifice plate is higher than other type of balanced flow meter because of the sudden contraction and expansion of the orifice. In this study, the genetic algorithm method is adopted as the optimization method to design a balanced flow meter using traditional orifice plate with enough pressure drop for instrumentation and small pressure loss for less pressurization requriements. In this thesis, we have combined the genetic algorithm with OpenFOAM (open source CFD code based on C++ programming language) and using balanced flow meter as an example of the design of optimization. We assume the flow is incompressible and Reynolds number Re=10,000 is used for the balanced flow meter. The two-equation turbulence model SST k-ω is adopted in this study. In this thesis, the pressure drop of 0.06 bar and 0.09 bar are set as the optimization target. The SUS (Stochastic Universal Sampling) method is used for selection and BLX-α is adopted for crossover. The results show that the optimized solution can be generally achieved after 10 iterations of genetic algorithm. This thesis is divided into four parts, which include the governing equations, benchmark of OpenFOAM fluid solver, genetic algorithm and discussion of optimization of the design of balanced flow meter.