Dynamic modeling and predictive control of a microfluidic system

TL;DR

This work addresses the challenge of regulating multiple flows in microfluidic chips under constraints. It develops a dynamic, physics-inspired model of the microfluidic system, then derives a linear reduced-order model for estimator and controller design. A Kalman-filter-based state observer and a model predictive controller are implemented and validated in simulation and on a real device, showing improved tracking and constraint satisfaction compared to PI controllers. The study highlights the impact of regulator dynamics and suggests online parameter adaptation as future work to enhance robustness and performance in changing operating conditions.

Abstract

Microfluidics, the study of fluids in microscopic channels, has led to important advances in fields as diverse as microelectronics, biotechnology and chemistry. Microfluidic research is primarily based on the use of microfluidic chips, low-cost devices that can be used to perform laboratory experiments using small amounts of fluid. These systems, however, require advanced control mechanisms in order to accurately achieve the flow rates and pressures required in the experiments. In this paper, we present the design of a model predictive controller intended to regulate the fluid flows in one of these systems. The results obtained, both through simulations and real experiments performed on the device, show that predictive control is an ideal technique to control these systems, especially taking into account all the existing constraints.

Figures (11)

• Figure 1: Schematic diagram of the microfluidic system used.
• Figure 2: Photograph of the microfluidic chip.
• Figure 3: Conceptual block diagram for one of the lines.
• Figure 4: Chip diagram
• Figure 5: Simulation results of the Kalman filter performance in the presence of noise.