Microfluidic networks can be applied to droplet-based Lab-on-a-Chip devices, where droplets are used to confine samples which flow through closed microchannels along different paths in order to execute (bio-)chemical experiments. In order to allow this routing of droplets, the design of the microfluidic network has to be precisely defined and afterward fabricated. However, neither the fabrication process nor the applied materials and components are perfect and, therefore, the fabricated microfluidic device frequently contains defects (produced by fabrication tolerances, properties of the used material, or fluctuation of supply pumps). Those may have a severe impact on the behavior of the microfluidic network and can even render the network useless. Furthermore, these defects complicate the design process, which eventually results in a “trial-and-error”-approach causing high costs with respect to time and money. Consequently, designers want to anticipate how robust their design is against those defects. This article, for the first time, describes how these defects can be abstracted, which eventually allows to evaluate the robustness already in the design process. We additionally introduce models considering single and multiple defects as well as corresponding methods for their analysis. Evaluations on a microfluidic network which is used to screen drug compounds confirm that the resulting robustness analysis indeed provides designers with a simple metric to decide how sensitive their design is against defects. The models and methods proposed in this article are grounded on the established 1-D analysis model.

中文翻译：

---

基于液滴的微流体网络的稳健性分析

微流体网络可应用于基于液滴的芯片实验室设备，其中液滴用于限制沿不同路径流过封闭微通道的样品，以执行（生物）化学实验。为了允许液滴的这种路由，必须精确定义微流体网络的设计，然后再制造。然而，无论是制造过程还是应用的材料和组件都不是完美的，因此，制造的微流体装置经常包含缺陷（由制造公差、所用材料的特性或供应泵的波动产生）。这些可能会对微流体网络的行为产生严重影响，甚至会使网络变得无用。此外，这些缺陷使设计过程复杂化，这最终导致“试错”方法，导致时间和金钱方面的高成本。因此，设计人员希望预测他们的设计对这些缺陷的鲁棒性。本文首次描述了如何抽象这些缺陷，最终可以评估设计过程中已经存在的稳健性。我们还介绍了考虑单个和多个缺陷的模型以及相应的分析方法。对用于筛选药物化合物的微流体网络的评估证实，由此产生的稳健性分析确实为设计人员提供了一个简单的指标，以确定他们的设计对缺陷的敏感程度。本文提出的模型和方法基于已建立的一维分析模型。