|dc.description.abstract||Microfluidics, the manipulation of nanoliter to microliter volumes of fluids, can give important new capabilities to researchers in biology, chemistry, material science, and medicine. Broadly, current methods can be classified into passive and active approaches. Passive methods use physical properties and microfluidic geometry whereas active methods use external perturbations to drive desired behaviour. However, passive methods require expertise and skill, and active methods complicate fabrication, require large support systems, or are not congruent with many applications. These limitations make microfluidics practically inaccessible to many researchers. Unlike these approaches, the application of feedback control may provide users with a practical and simple way to use microfluidics. Through feedback, a controller manages the operation of a microfluidic chip without needing complicated fabrication, large support systems, and in a way that can be used in a wider set of applications.
State-of-the-art feedback-controlled microfluidic (FCM) devices have several shortcomings. First, typical microfluidic chips used in these devices are simple single or double T-Junctions. Such simple chips have few degrees of freedom thus limiting how many channels can be controlled concurrently. Secondly, feedback techniques, predominantly based on optical microscopes, are bulky and costly thus incongruent with FCM applications which require compact and low cost sensing. Third, a modeling approach based on an electrical analogy (Modeling channels through resistive, capacitive, and inductive elements) leads to untenable models. Fourth, Linear Quadratic Regulator (LQR) based control laws saturate unidirectional pumps. Finally, current FCM methods necessitate significant operator interaction, which is undesirable.
To improve FCM methods, this dissertation conceives a new type of chip topology with greater degrees of freedom. Secondly, a new feedback source based on lensless microscopy is developed and validated. Third, a simplified modeling approach is validated. The simplified model is used as the basis for a Model Predictive Controller (MPC). Finally, these subsystems are combined to develop a system that can generate and manipulate droplets autonomously. These developments work towards making FCM, and thus microfluidics, more accessible to the wider scientific community.||en