Research On The Integrated Design Method Of Microfluidic Channel Networks On Chips

The advantages of microfluidic networks, such as efficiency, portability,phenomena of micro scale flow, have made it utilized more and more popularespecially in the biochemistry analysis. However, the limitations in design methodand processing technology are surely hampers for the development of high accuracyintegrated design. Based on the method of topology optimization, this thesis focuseson the research on the optimization design of microfluidic networks, especially theones related to controlling the flow rates on the outlets. This method is validated andwould be extended to the components’ design work related to the flow rates anddirections controlling.Flow rates distribution and direction control are the two key components for thedesign of a microfluidic channel network. What we have to do is to design a networkwhich guides the distribution of fluid with specified flow rate into specified channels.The most popular method to accomplish this design is the electrical analogies method.Other methods, such as trial and error method and constructal theory and so on, arealso utilized in microfluidics. With the rapid development of the microfluidic, thesemethods cannot hit the requirements like large scale, high integration andhigh accuracy. Whereas, the numerical design method catches the attention for itsconvenient and fast characteristics. The numerical method is based on physics lawsand mathematical derivation, what make it more reasonable.In this thesis a method based on topology optimization to design microfluidicespecially the networks with specified outlet flow rates is proposed.(1) research onthe design method of the microfluidic especially the electrical analogies method isdone. Merits and demerits are summarized based on previous researches.(2) a newway is proposed to design microfluidic networks based on topology optimizationutilizing the method of “Real Reference” models. This method is proved to improve the speed of convergence and decrease the difficulty to set the models.(3) the methodof calculating sensitivities is discussed when the calculating domain different with thedesign domain.(4) a new implementation of the constraint of the channel volume isproposed which makes the optimization more reasonable and stable.(5) codes forpreprocessing and postprocessing are programmed in order to make this method moreconvenient for new users. Examples are done via the method of CG, MMA, LBFGS totest and verify this method.(6) different optimization models, like models with oneinlet and multi outlets, models with multi inlets and multi outlets, cascade design andso on, are optimized designed. Numerical results are compared in order todemonstrate the performance of the optimization method.(7) this method is extendedto design components relating to the flow rate and flow directions. The concentrationgradient networks are detailed researched and verified.Finally the advantages and the points need to be improved of the proposednumerical optimization method are discussed. Combined with the electric analogiesmethod, it would promise to be a supplement method in design microfluidic networks.