Tejas Patel

Hi, welcome to my personal CFD gallery. Professionally, I have over 7 years of experience in researching diverse fluid flows. My competence lies in studying thermal & physical fluid flow problems. I am also proficient in the post-analysis and development of CFD softwares / numerical codes.

At present, I am a CFD Research Engineer at Convergent Science Inc. in Madison, Wisconsin. Before this, I was a Ph.D. student in the Mechanical Engineering department at Michigan State University where I worked conjointly as a researcher in the Computational Biomechanics Lab under Dr. Lik Chuan Lee and the Complex Fluids Lab under Dr. Tong Gao.

What interests me? Fluid structure interactions, aerodynamics/aeroelasticity of vehicle/bikes components, wind-tunnel testing, flow turbulence, shallow-water waves and bio-physical, multiphase and compressible flows.

Enjoy my CFD archive and feel free to reach out with job opportunities, consultation or if you want to simply talk fluids.

Let's connect:

WHAT IS CFD ?

Where there is fluid, there is CFD. CFD is short for Computational Fluid Dynamics. It is a branch of fluid mechanics that uses numerical analysis to produce quantitative predictions of fluid-flow phenomenon based on fundamental conservation laws.

A CFD analysis has great potential to save time in the design process and are therefore cheaper and faster compared to conventional experimental testing. For predicting the flow behavior in a product or a physical situation, a simulation is normally conducted under a set of assumed geometry, boundary conditions and initial states. CFD (as a tool) has improved in the last decade, however its predictions are never completely exact. Even today, despite the use of high-speed supercomputers, some large-scale simulations take over a week to produce results within reasonable accuracy.

CFD GALLERY

1. Cryoballoon-ablation for Pulmonary Vein Isolation

Cryoballoon Ablation (CBA), a recent cryo-energy based invasive technique, is an established surgical procedure for patients suffering from Atrial Fibrillation (AF). In an AF patient, irregular electrical signals usually begin where the Pulmonary Veins (PV) attach to the Left Atrium (LA). In such a patient, it becomes very important to cease the transmission of irregular electrical activity from the PV to the heart chambers. During CBA, the electrophysiologist positions a cryoballoon at the PV ostium and then liquid nitrous oxide is injected in the cryoballoon to form a circumferential lesion. Overall CBA is safer, more predictable, forms a contiguous lesion and is less time consuming than radiofrequency ablation, which is the more traditional approach.

For a successful CBA surgery, important factors controlling lesion formation are balloon-tissue contact area, time of freeze and the source of cryo-energy. It is also very critical to montior real time temperature distribuiton in the neighbouring non-cardiac tissues to avoid serious thermal complications, i.e. phrenic nerve injury and esophageal injury. Despite the existing state of art for CBA, thermal distribution in the pulmonary vein antrum is poorly understood and using simulations can shed light to predict the desired efficacy. At Michigan State, I developed a stabilized finite-element formulation in FEniCS to simulate CBA in the PV antrum on a patient-specific LA geometry. The model accurately predicts lesion size, transfer of cryo-energy and hemodynamics during CBA. Different cryoballoon positions are simulated; this helps the electrophysiologist to optimize the treatment.

2. Multiphase flow using Immersed-Boundary Method (IBM)

Multiphase flow refers to the simultaneous flow of more than one fluid phase. They are ubiquitous and modeling such flows require accurate tracking of the fluid-fluid interface. The large majority of processing technology involves multiphase flow. The most common class of two-phase multiphase flows include Gas-Liquid flow, Gas-Solid flow, Liquid-Liquid flow and Liquid-Solid flow. Some widely studied examples include: sediment transport in rivers, flow of RBC in blood plasma, water electroysis, harvesting tidal energy and carbon-dioxide sequestration.

I have developed a novel numerical method based on a Dual-Grid Dual Level Set function to simulate Gas-Liquid & Liquid-Liquid multiphase flows in complex geometries. A Finite-volume/Finite-difference code is developed, which employs the Cut Cell based IBM. The multi-phase boundary is simulated using the DGLSM and the immersed stationary boundary is modeled using a secondary level set function. The solver can cope up with break-up, distortion, and re-combination of the interface, even for high Re within 5% mass accuracy. Above, a sample dam-break problem shows the transfer of potential energy into kinetic energy as it runs into the immersed boundary. Another problem shows the development of the physical interface for Kelvin-Helmholtz instability. This is normally observed in atmospheric cloud layers and in oceans where there is density difference at higher sea depths.

3. FEM based Fluid-Structure Interaction

Fluid–Structure Interaction (FSI) is the interaction of a movable or deformable structure with an internal or surrounding fluid flow. Such interactions can be stable or oscillatory. FSI modeling is highly challenging, especially when the objects are thin and include large-scale deformations. Some common engineering applications involving FSI are predicting aeroelastic vibrations, design of parachutes, flow in microfluidic devices, biomechanics of urinary bladder & heart chamber and studying arterial stenosis & aneurysms. At Michigan State, we strive to study the biophysical modeling in active cardiac chambers to better investigate cardiac diseases and conditions.

I have developed a Fictitious Domain (FD) algorithm using stabilized Finite Element Method (FEM) to perform large-scale biophysical FSI simulations with heat transfer. The code effectively scales well up to 40 mi elements on high performance clusters. The key idea of the FD method is to virtually extend the fluid domain into the solid, where a distribution of pseudo body forces is employed to enforce structural movement through Lagrange multipliers. The fluid is solved on a fixed Eulerain grid whereas the solid deformation is computed on a flexible lagrangian grid. This significantly lowers computational expense by avoiding temporal remeshing. As shown here, physical benchmark validations are performed to validate the solver.

4. Turbulance modelling using OpenFOAM

A turbulent flow field is characterized by velocity fluctuations in all directions and is chaotic, diffusive, dissipative, and intermittent. The most important characteristic of a turbulent flow is the infinite number of scales such that a full numerical resolution of the flow (i.e. a DNS), requires the construction of a grid with a number of nodes that is proportional to Re ^9/4 . So to compute such turbulance, we decompose any variable as a sum of its average and the fluctuations. The new equations will be exact for an average flow field (such that the property becomes constant over time) but not exact for the turbulent flow field. Using such a decomposition results in the RANS or Reynolds Averaged Navier Stokes Equations.

In the RANS decomposition, new unknowns are introduced such as the turbulent stresses and turbulent fluxes which require modelling. Some widely used models are the two-equation models and the Reynolds Stress Model (RSM). A backward facing step flow at Re ~ 56000 is simulated using pisoFOAM. The two equation models (or eddy viscosity models: K-epsilon, K-omega model) are first order accurate for which the solver computes turbulant kinetic energy and turbulent dissipation. On the other hand, the RSM model solves seperate equations for turbulent stresses and fluxes and is second order accurate, however computationally it is most expensive. Above, one can see that transient vortices are captured better using the RSM model, than the eddy viscosity models. At steady state, the K-omega and RSM model predict approximately similar velocity profiles and vortex strength.

5. Bubble-rise dynamics in corrugated channels

The study of gas bubble in corrugated channels is of great importance because they are relevant in several applications including bubble column reactors, cavitation, heat exchangers and nuclear reactors, to name a few. Early attempts to study bubble phenomena were restricted to experiments. The fate of such bubbles and their physics and is largely dependent on the Morton number (ie. bubble properties), adjacent fluid properties and the channel confinement ratio. Despite prologned efforts, even today the accurate interpretation of bubble behavior poses a challenge due to the nonlinearity and 3D nature of the problem.

In our Finite-Volume computational model, a spherical bubble is left free to rise in a corrugated channel: defined using a sinusoidal function. First, validatations are matched with previous experimental findings. Next, we study the effect of corrugation on the bubble physics. We find that these rising bubbles adhere to a characteristic shape in each regime: defined based on Bond number and Reynolds number. Path instability (wobbling) is normally observed at higher Reynolds number. On the contrary, skirt formation is detected at higher Bond numbers. As shown in the figure, with an increase in channel wall amplitude - we observe that secondary vortices are formed in the wake, wobbling is suppressed, drag increases and the average rise velocity drops.

Education

Doctor of Philosophy,  Mechanical Engineering |   2023  

Michigan State University, East Lansing, MI
GPA: 3.81/4

Bachelor of Technology,  Mechanical Engineering |  2017
Minor,  Design Engineering

Nirma University, Ahmedabad, India
GPA: 8.48/10

Graduate Courses

Continuum mechanics, Cardiovascular mechanics, Linear elasticity, Thermoelasticity and Viscoelasticity

Inviscid flow, Complex fluids, Compressible flows (Gas dynamics), Aerodynamics and Aircraft performance, Fluid structure interactions, Turbulence modelling

Finite element method, Parallel computing, Numerical Linear Algebra

Technical Skills

Solidworks, OpenFoam, COMSOL MultiPhysics, ANSYS : fluent | structural | thermal, FEniCS, HyperMesh, Altair Inspire, Gmsh, Paraview, Tecplot

Finite Element Method (FEM), Finite Volume Method (FVM), Finite Difference Method (FDM)

C, C++, Python, MATLAB

Linux, High Performance Computing (HPC), MS Powerpoint

Publications

1. Patel, Tejas, Chris Li, Farshad Raissi, Ghassan S Kassab, Tong Gao, Lik Chuan Lee. "Coupled Thermal-Hemodynamics Computational Modeling of Cryoballoon Ablation for Pulmonary Vein Isolation." Computers in Biology and Medicine 157 (2023): 106766.
(Access article here)

2. Patel, Tejas, Darshan Patel, Nihar Thakkar, and Absar Lakdawala. "A numerical study on bubble dynamics in sinusoidal channels." Physics of Fluids 31, no. 5 (2019): 052103.
(Access article here)

3. Patel, Tejas, and Absar Lakdawala. "A dual grid, dual level set based cut cell immersed boundary approach for simulation of multi-phase flow." Chemical Engineering Science 177 (2018): 180-194.
(Access article here)

Fellowships & Awards

1. Secured 2^nd place in Ph.D. Podium Competition at Summer Biomechanics, Bioengineering and Biotransport Conference (SB3C) - 2022.
Access my Certificate

2. Graduate Office Fellowship (GOF) for summer 2022 awarded by College of Engineering at Michigan State University.

3. Member at American Society of Mechanical Engineers (ASME) - 2023.
Access my membership certificate