針對不同應用的微流體最佳化

Optimization of Microfluidic Channels for Different Applications

李強(Imtiyaz Hussain)

指導教授 ： 陳致真

國立清華大學/工學院/動力機械工程學系/碩士(2021年)

摘要

出於模擬計算時長及電腦儲存容量上的限制，涉及流體流動及其它參數的微流道最佳化繁瑣且不實際。與3D數值模擬相比，使用經驗式或近似解能縮短模擬計算時長。並且，經驗式可以預先將相關的設計及處理參數考慮進去，因此本研究大量使用之，結合最佳化技術來找到最適當的設計及處理參數變的可行。若給定的參數較少，還可以設定許多可調的參數。這項研究開發了一種包含微流體五種不同應用的軟體。此研究強調了五種應用的最佳化:單一流體、物質的混合和稀釋、蛇形通道中的快速混合、混合與反應 I（一混合就會反應）以及混合與反應 II（混合完成後才開始反應）。最佳化是基於晶片限制（晶片面積）和一些給定的參數，如流速、混合效率等等。這些參數由統御方程式控制，如流體流動的 Navier-Stokes 方程式、彎曲通道的 Dean 流動、混合和稀釋的擴散方程式以及欲達到的產率。本研究使用不同的最佳化演算法，如迭代計算優化、基因演算法、雙重退火和差分進化。最佳化後的或給定的微流道長度會自動擬合在一個或多個入口和出口之間，並使用海龜繪圖系統繪製最佳化後的晶片。模擬結果藉由 Comsol 和文獻中的其他解析方程式完成驗證。用戶界面內置於 Pyqt5 和 PySide 設計器中。該模型的主要優點是與其他模擬軟體相比，內置應用程式的計算時間非常短（ < 15 分鐘）。

關鍵字

最佳化 ； 基因演算法 ； 迭代計算優化 ； 雙重退火 ； 差分進化 ； Pyqt5 和 PySide Designer ； 相關性 ； 3D 數值實現

並列摘要

Due to computational time and computer memory limits, optimizing microfluidic channels involving fluidic flow and other processes becomes cumbersome and infeasible. The computational time is reduced by considering correlations (analytical equations) or numerical approximations compared to the 3D numerical implementation. Also, process parameters prior to design parameters can be considered if correlations are used. Since the correlations are used, optimization techniques can be incorporated to find the optimal value of process or design parameters and many optimal parameters if fewer parameters are given. This study has developed the software, which contains five different applications used in microfluidics. The proposed research highlights the optimization of five applications: microfluidic channels containing single fluid, mixing and dilution, rapid mixing in serpentine channels, mixing and reaction I (reaction occurs as soon as the mixing occurs), and mixing and reaction II (reaction starts after complete mixing). The optimization is based on chip constraints (area of the chip) and some input dimensions such as flow rate, the efficiency of mixing, etc., which are governed by fundamental equations such as Navier-Stokes equation for fluid flow, dean flow for curved channels, diffusion equation for mixing and dilution and also the rate of removal if a reaction occurs. This study uses different optimization algorithms such as iterative optimization, genetic algorithms, dual annealing, and differential evolution. The optimized or given length is automatically fitted between the inlets and the outlets, and turtle graphics draw the optimized chip. The validation of results is done in Comsol and other analytical equations found in the literature. The user interface is built in Pyqt5 and PySide designer. The primary advantage of this model is that the computational time is very short (<15 min) for built-in applications as compared to other simulation software (usually greater than 15 min).

並列關鍵字

Optimization ； Genetic Algorithms ； Iterative Optimization ； Dual Annealing ； Differential Evolution ； Pyqt5 and PySide Designer ； Correlations ； 3D Numerical Implementation

參考文獻

1. Bruus, H., Theoretical Microfluidics. Department of Micro and Nanotechnology Technical University of Denmark. 2008, Oxford, New York.

2. Figeys, D. and D. Pinto, Lab-on-a-chip: a revolution in biological and medical sciences. 2000, ACS Publications.

3. Curtis, A. and C. Wilkinson, Nantotechniques and approaches in biotechnology. Trends in Biotechnology, 2001. 19(3): p. 97-101.

4. Clayton, J., Go with the microflow. Nature Methods, 2005. 2(8): p. 621-627.

5. Becker, H. and L.E. Locascio, Polymer microfluidic devices. Talanta, 2002. 56(2): p. 267-287.