Expanding the Microfluidic Design Automation Capabilities of Manifold: Electrophoretic Cross and Time-Domain Simulation

Abstract

Lab-on-a-chip devices are finding applications in several different fields, from point-of-care diagnostics to genome sequencing. However, lab-on-a-chip is a multidimensional field that makes it difficult for designers to have a full understanding of the entire system. There currently lacks a computer aided design (CAD) tool that allows microfluidic designers to express partial designs, only defining the parts of the system that they know and the tool determines the rest of the system while still ensuring the device will operate as expected. This results in devices being tested by physically constructing them and performing multiple design iterations should the prototype fail to operate correctly, increasing the time and cost of microfluidic design. The Manifold language was developed to address this problem by allowing the microfluidic designer to specify the parameters that they know and then Manifold solves for the ranges that the rest of the parameters can take, reducing the cognitive load required to design microfluidic devices. This thesis discusses the improvements that were made to Manifold's design capabilities to create Manifold V3.0: the addition of electrophoretic cross channel simulation and the ability to simulate designs in the time-domain in MapleSim through the use of Modelica. The Modelica design is generated automatically, creating a feedback loop that allows the designer to see their microfluidic device in operation before manufacturing a prototype. Finally, a preliminary validation of the software was performed through the comparison of Manifold's simulations to historical data collected from real microfluidic devices. This validation was structured as seven research questions that are asked of Manifold and they are each worked through using the historical data to determine if Manifold is able to answer these questions.