This paper is devoted to developing discrete microfluidics using multidimensional network models (MNM). The method provides the possibility of generating a set of possible variants of the invention in the form of MNMs and selecting the best design variant using analyzing both the structure and the features of the MNMs. For example, the paper reports results of designing the microfluidic three-stage generator, including the design MNM and the experimental model.