Session T18: Microfluidic Flow Applications I

Physics Informed Neural Network based Modelling of Microchannel Flow Dynamics

The flow inside microchannels is experimentally studied using Particle Image Velocimetry (PIV) and numerically using computational fluid dynamics (CFD). However, depending on the model used for simulation, there is often a mismatch between the experimental data and the numerical counterpart. We devise a technique based on Physics-Informed Neural Networks (PINN)  to solve the momentum equations by incorporating some data points from the experiments to get the most accurate model of the flow device/ channel. The loss function of the neural network is modified as per Navier-Stokes equations, and the continuity equation. The physical system modeling is done for the microchannel dimensions, for which computational cost is prohibitive at times, and experiments are not always feasible. We compare the effectiveness of our PINN model with CFD techniques to provide a robust alternative addressing experimental uncertainties. Such methodologies are effective in modeling the flow inside microdevices that have multifaceted implications in biosensing, drug discovery and delivery, point-of-care testing, and future energy devices.   

Using Theory, Simulation, and Experiment to Probe the Multimodal Thermal Noise Spectrum of a Nanobeam in Fluid Near a Wall

Technology continues to push towards smaller objects to develop increasingly sensitive sensors. By immersing these sensors in a fluid, the resonant frequencies and quality factors shift, and the presence of a nearby wall influences the fluid-solid interaction. To understand these differences, we examine the dynamics of a doubly-clamped beam under high tension in air or water. We explore the Brownian noise spectrum of the beam displacements using theory, simulation, and experiment. We provide a deterministic theoretical description using the fluctuation-dissipation theorem, where the dissipation is due to the viscous fluid. Additionally, we use a finite element approach to study the 3D fluid-solid interaction between the elastic beam and the surrounding fluid. We compare our simulations and experiments using different models of the hydrodynamic function to investigate the effects of a nearby wall on the thermal noise spectrum. We examine the first eleven modes of the noise spectrum, finding excellent agreement between theory, simulation, and experiment. We find that a nearby wall influences the noise spectrum for frequencies with a corresponding Stokes' length greater than 1/5 of the separation between the wall and the beam for the parameters we explore.   

Diffusion into microchannels with dead-end pores of non-uniform cross-sections

Understanding microscale fluid flows is critical to perfecting the manufacturing and use of microfluidic technologies for medical and engineering applications. Microchannels with dead-end pores are ubiquitous in natural and industrial settings, and ongoing research focuses on solute and particle transport in and out of these pores. We present a repeatable and accessible experimental protocol developed to study the passive diffusion of a dissolved solute into microchannels with dead-end pores of rectangular as well as widening and narrowing trapezoidal geometries. Custom microchannels with pores of specified domains are produced using a rapid-prototyping technique involving inexpensive materials and a commercial craft cutter. The experimental data is compared directly to 3D numerical simulations as well as analytical solutions of a modified 1D diffusion model: the Fick-Jacobs equation. The role of pore geometry on the passive diffusion process will be highlighted. Ongoing and future directions will be discussed.   

Sensitivity of the Deposition Rate to the Surface Reaction Parameters during Chemical Vapor Infiltration of SiC Matrix Composites

Chemical vapor infiltration (CVI) is typically used to fabricate high-purity matrix composites for extreme environments. In CVI, a gas-phase precursor is injected into a chamber, which diffuses and reacts on the surface of a fiber preform. The deposition quality and rate depend upon the surface reactions, which can be controlled in different ways, such as higher operating temperature, thermal-gradient CVI, forced-flow CVI, etc. A computational model is desirable for parametric studies, but such an approach should accurately describe transport, energy interactions, interface dynamics, and surface reactions occurring during CVI. In this study, we employ the Direct Simulation Monte Carlo (DSMC) technique to simulate the flow around microscale fibers. DSMC is a robust and efficient tool for solving rarefied gas dynamics using a molecular kinetic scheme with a particle-based probabilistic approach. It allows examining fundamental aspects of molecular motion, intermolecular collisions, and chemical reactions. The sticking coefficient of the chemical species participating in the surface reactions affects the deposition during CVI. It depends upon parameters such as the preform structure, surface coverage, and operating conditions. It is usually determined empirically from experiments. In this study, we examine the sensitivity of the deposition rate of SiC to the sticking coefficients of chemical species in fiber preforms of different sizes (1 – 10 μm) and at a range of operating conditions.   

Propeller can't propel at Intermediate Reynolds Numbers- Experiment

Microrobots hold immense potential for a wide range of applications, including targeted drug delivery, vascular plaque removal surgery, and various industrial fields. Enhancing microrobot performance often focuses on designing propellers across millimeter, micrometer, and nanometer scales. At the micrometer and nanometer scales, these propellers are frequently designed as helical structures. Extensive research has also been conducted on propellers operating at high Reynolds numbers (Re) in aviation. However, there is a notable lack of research on propeller performance at intermediate Re.

In this study, a propeller-driven toy submarine was used to examine thrust performance in silicone oil with varying viscosities to adjust Re. Our results show that for Re ranging from 5 to 130, the toy submarine consistently moves in reverse, regardless of the propeller's rotation direction. By modeling the submarine as a disk-propeller combination, we confirmed that at intermediate Re conditions, the interaction between the submarine body and the propeller causes the thrust direction to reverse even when the propeller rotates forward. Additionally, we observed that the range of Re where this reversal occurs widens as the disk diameter increases or as the disk is positioned closer to the propeller.   

Unified mobility expressions for externally driven and self-phoretic propulsion of particles

Technologies involving artificial micro-swimmers are advancing for targeted drug delivery, diagnostics, and environmental cleanup, yet face unique challenges in-vivo compared to controlled in-vitro environments. Understanding microswimmer propulsion across different conditions is crucial. The mobility of external and self-propulsion of particles is evaluated by simultaneously solving the solute conservation equation, interaction potential equation, and the Stokes equation with a body force. This method, though accurate, becomes complex, especially at finite interaction length scales. Inspired by Brady JFM (2021), we obtain unified mobility expressions with arbitrary interaction potentials. Firstly, we show that these expressions can recover well-known mobility relationships in external electrophoresis and diffusiophoresis for arbitrary double-layer thickness. Secondly, at the thin interaction length limit, these equations reduce to the slip velocity expressions for spherical microswimmers well-known in active particle literature. Finally, we explore the dynamics of autophoretic ellipsoidal Janus particles, investigating how particle eccentricity, surface heterogeneities, external gradients, and interaction length scales impact their swimming speeds.   

Short- and long-term dynamic modelling of pressure-driven flow for droplet microfluidic applications

Microfluidic devices are commonly driven by syringe or pressure pumps. Syringe pumps cause persistent flow oscillations and have slow response times. Comparatively, pressure pumps typically offer stable output and a fast response time. However, the compressed air provided by the pressure pump must be interfaced with the liquid samples at the so-called reservoir holder. The multi-phase system introduces short- and long-term dynamics. Long-term dynamics are of interest for passive microfluidic systems while active microfluidic systems benefit from modelling short-term dynamics for consideration in the controller design. The lack of quantified analysis limits the performance of these systems. The model herein proposed starts from the fundamental principles (conservation of mass, conservation of energy). Key assumptions enable the formulation of a system of differential equations describing the short- and long-term dynamics. Both a passive and an active system are considered as case studies with varying parameters such as reservoir volume and shape, and microfluidic chip resistance. The analysis allows better-informed decisions when designing a microfluidic system as well as uncertainty estimation for experimental data collection.   

From hydrodynamics to clogging in emitters used in drip irrigation

Drip irrigation is a water-saving technology that also increases the efficiency of agricultural production. However, in many cases, the adoption of drip irrigation is limited by clogging of the drip emitters through particle deposition or biofouling. Indeed, a geometry of drip emitter widely used in agriculture consists of a labyrinth channel of millimetric dimensions in which the Reynolds number is in the range 500-1000. Particles can deposit at different locations in this channel and eventually clog the entire system over some time. Here, we consider the flow profiles in 3D-printed prototypes of drip emitters using PIV methods and numerical simulations. We also conduct experiments of particle deposition in these channels to identify the preferential location of particle clogging and connect these sites with the local hydrodynamic. This approach allows us to tune the geometric features of the labyrinth channel to decrease particle deposition and thus increase the lifetime of the drip emitter.   

Dynamics of Flexible Microfibers in Shear and Vortical Flows

At the microscale, it has been argued that phytoplankton such as diatom chains experience turbulence locally as a linear shear flow. However, the lack of vorticity gradients and vorticity transport in such interpretations can lead to oversimplification. In this work, we study the shape evolution of slender flexible fibers in different external flows. Our computational framework is a Kirchhoff rod model coupled to regularized Stokeslet segments. Previous experimental and numerical work demonstrated morphological transitions of passive elastic fibers from tumbling to S-turns to snaking in shear depending on a nondimensional elastoviscous number. We first validate our model by capturing these behaviors in shear and then transition to an analog of a Burgers vortex at zero Reynolds number.  This flow is created by the superposition of several regularized singularities of the Stokes equations. We will present model results that exhibit rich shape dynamics and excursions of flexible microfibers in these vortical flows, and will also discuss how these depend on an appropriately defined elastoviscous number.