arXiv subjects

Arezoo Ardekani

Publications and source records attributed to Arezoo Ardekani.

3 recordsLinked to original sources

arXiv2026

Turbulent oscillation in unbalanced T-junction flows

The T-junction impinging flow occurs in many fluid dynamics systems. In particular, the T-junction micromixer has recently been widely used for nanoparticle production, where the two inlet streams operate at a significant flow-rate imbalance and the Reynolds number is in the turbulent regime. This operating condition exposes a gap in the existing literature on the fluid dynamics of the T-junction. In this study, we used high-fidelity numerical simulations to investigate high-Reynolds-number unbalanced T-junction flows. We discover a new oscillatory behavior between the two inlet streams at the T-junction, leading to a new turbulence-production mode. We will present detailed evidence of this new behavior, in contrast to the existing understanding of balanced turbulent T-junction flows. This oscillatory behavior also persists across a range of Reynolds numbers simulated, where the Strouhal number is approximately constant, indicating a self-similar phenomenon. As a result, many of the fluid dynamics parameters follow a power-law relation with the Reynold number. The discovery in this paper affects real-world applications, where process design and product quality are affected by turbulence and mixing dynamics.

physics.flu-dyn↗

arXiv2026

High-Fidelity Simulations of Two Miscible Fluids in Small Scale Turbulent Mixers Using a Variational Multiscale Finite Element Method

Turbulent mixers have been widely used in industrial settings for chemical production and increasingly for therapeutic nanoparticle formulation by antisolvent precipitation. The quality of the product is closely related to the fluid and mixing dynamics inside the mixers. Due to the rapid time scales and small sizes of many turbulent mixing geometries, computational fluid dynamics simulations have been the primary tool used to predict and understand fluid behavior within these mixers. In this study, we used the residual-based variational multiscale finite element method to perform high-fidelity turbulent simulations on two commonly used turbulent mixers: the multi-inlet vortex mixer (MIVM) and the confined impinging jets mixer (CIJM). We simulated two geometric variations, two-inlets and four-inlets, of the MIVM and two different inflow ratios of the CIJM. Through detailed turbulence results, we identify differences in turbulence onset, total energy, and mixing performance of the two MIVM configurations. With the CIJM results, we demonstrate the effect of the flow rate ratio on the impingement behavior, and as a result, on the mixing performance and turbulence. The cross-comparison between the two mixers shows key differences in turbulence and mixing behaviors, such as the turbulence onset, the energy decay, and the output mixing index. This study demonstrates the importance of a high-accuracy numerical scheme in simulating the turbulent mixers and understanding the similarities and differences among mixers. Furthermore, the results show potential for optimizing the operating conditions to achieve the best mixing performance.

physics.flu-dyn↗

arXiv2023

Temporal Consistency Loss for Physics-Informed Neural Networks

Physics-informed neural networks (PINNs) have been widely used to solve partial differential equations in a forward and inverse manner using deep neural networks. However, training these networks can be challenging for multiscale problems. While statistical methods can be employed to scale the regression loss on data, it is generally challenging to scale the loss terms for equations. This paper proposes a method for scaling the mean squared loss terms in the objective function used to train PINNs. Instead of using automatic differentiation to calculate the temporal derivative, we use backward Euler discretization. This provides us with a scaling term for the equations. In this work, we consider the two and three-dimensional Navier-Stokes equations and determine the kinematic viscosity using the spatio-temporal data on the velocity and pressure fields. We first consider numerical datasets to test our method. We test the sensitivity of our method to the time step size, the number of timesteps, noise in the data, and spatial resolution. Finally, we use the velocity field obtained using Particle Image Velocimetry (PIV) experiments to generate a reference pressure field. We then test our framework using the velocity and reference pressure field.

physics.flu-dyn↗