Research

Please see below for a short summary of my ongoing and previous research projects. For some of the projects, I cannot provide complete details at this moment or ever due to confidentiality.

A Comparative Study of Turbulence Models for Two-Phase Coaxial Swirling Jet Flows

Status: Complete

Funding agency (Role): Gas Turbine Research Establishment - DRDO (as a postdoctoral researcher at IIT Madras)

Summary: This study assesses different turbulence modeling approaches for simulation of two-phase coaxial annular swirling jet flows. The problem selected from literature involves an analytical inlet profile for an annular liquid sheet sandwiched between two coaxial annular gaseous jets. The liquid-gas interface is resolved using the volume-of-fluid (VOF) model with continuum surface force approximation. 3D unsteady Reynolds averaged Navier-Stokes simulations using up to 8 million grid cells and 64 HPC cores are conducted to obtain transient multiphase CFD data for this case. Different turbulence models explored include the k-epsilon RNG with swirl modification, the Reynolds stress model (RSM) and RSM with scale adaptive simulations (RSM-SAS). Comparisons with the direct numerical results from literature suggest that the simulation using RSM-SAS approach better predicts the onset of instability, liquid jet column collapse, jet mixing, vortex breakup, and the overall characteristics of this flow.

Relevant publications: 

1. Choudhary A, Narasimhamurthy VD. A Comparative Study of Turbulence Models for Two-Phase Coaxial Swirling Jet Flows. In: 25th International Conference on High Performance Computing Workshops (HiPCW). Bengaluru, India: IEEE; 2018. doi:10.1109/HiPCW.2018.8634062.

2. Choudhary A, Narasimhamurthy VD. DES and RANS Modeling of Primary Atomization In a Coaxial Swirling Liquid-Gas Jet. Atomization and Sprays. 2023; 33(5):47-74. doi:10.1615/AtomizSpr.2023045729

3. Asapu S, Pandey A, Choudhary A, Sahu S, Narasimhamurthy VD, Two Phase Flow Simulation of Primary Liquid Breakup in Coaxial Jets, 8th International and 47th National Conference on Fluid Mechanics and Fluid Power (FMFP), Assam, India; 2020.

Numerical Error Estimation for Fluidized Bed Simulations

Status: Complete

Funding agency (Role): Oak Ridge Institute for Science and Education (ORISE) (in part, while working as a postdoctoral fellow)

Summary: Numerical error estimation (or solution verification) is the process of assessing the accuracy of a numerical solution by identifying and estimating different types of numerical errors present in the computational simulation. Various sources of numerical errors include discretization error (i.e., error due to grid size or time-step), iterative error (i.e., error due to insufficient iterative tolerance), and round-off error (i.e., error due to machine precision). Assessment of numerical accuracy of a System Response Quantity (SRQ) in a computational simulation is an important step before using the simulation results for performing validation against experimental results or for engineering prediction. 

In this work, the goal is to present numerical error estimation for two-fluid simulations performed using the multiphase flow code MFIX (Multiphase Flow with Interphase eXchanges). The Two-Fluid Model (TFM) of MFIX is employed in which the different phases (e.g., a gas and a solid phase) are treated as interpenetrating continua. The interactions between the different phases are modeled using various interphase exchange terms (or exchange models) such as gas-solids drag model, solids-solids drag model, solids-solids stress model, gas-solids heat transfer model. There are several challenges in performing V&V activities for fluidized bed systems that this project aims to address. These challenges are (a) inherent unsteadiness of the gasification process, (b) large computational expense requiring large time-steps and iterative tolerances on the engineer's part, and (c) effects due to grid refinement that are not seen in simpler, single-phase flow systems.

Relevant publications: 

1. Choudhary A, Roy CJ. In: Yeoh GH, editor. Verification and Validation for Multiphase Flows. Singapore: Springer; 2018. p. 1–37. doi:10.1007/978-981-4585-86-6_24-1.

Residual-based Mesh Adaptation on Structured Grids

Status: Complete

Funding agency (Role): US Air Force Office of Scientific Research (AFOSR) (in part, while working as a GRA at Virginia Tech)

Summary: In a CFD simulation, truncation error (TE) is the difference between the continuous governing equation and its discrete approximation. Since TE can be shown to be the local source term for the discretization error, TE is proposed as the criterion for determining which regions of the computational mesh should be refined/coarsened. For mesh modification, an error equidistribution strategy to perform r-refinement (i.e., mesh node relocation) is employed.  The proposed technique is applied to 1D and 2D flow problems for which exact (i.e., analytic) solutions are available.  For mesh adaptation based upon TE, about an order of magnitude improvement in discretization error levels is observed when compared with the uniform mesh. For a simpler problem (1D Burgers equation), we find TE-based adaptation to give better results compared to feature-based adaptation (e.g., solution gradient, solution curvature). 

Relevant publications: 

1. Choudhary A, Tyson WC, Roy CJ. Implementation and Verification of a Residual-based r-Adaptation Technique On Structured Meshes. ASME Journal of Verification, Validation and Uncertainty Quantification. 2019 April; [ACCEPTED]

Papers in Conference Proceedings

1. Tyson WC, Derlaga JM, Roy CJ, Choudhary A. Comparison of r-Adaptation Techniques for 2-D CFD Applications. In: 22nd AIAA Computational Fluid Dynamics Conference. Dallas, Texas: American Institute of Aeronautics and Astronautics; 2015. doi:10.2514/6.2015-2611. 

2. Choudhary A, Roy CJ. Structured Mesh r-Refinement using Truncation Error Equidistribution for 1D and 2D Euler Problems. In: 21st AIAA Computational Fluid Dynamics Conference, Fluid Dynamics and Co-located Conferences. American Institute of Aeronautics and Astronautics; 2013. doi:10.2514/6.2013-2444. 

Probability Bounds Analysis Applied to Sandia V&V Challenge Problem

Status: Complete

Funding agency (Role): N/A (while working as a GRA at Virginia Tech)

Summary: In 2014, Sandia National Labs invited teams from various organizations to participate in a Verification and Validation workshop. In the (hypothetical) problem, a corporation maintains a large number of storage tanks that hold a  liquid under pressure. This loading causes deformations of the tank walls. During a standard safety inspection, measurements on one tank violated the safety specifications. The goal of each participating team would be to develop and execute an analysis strategy using FEM simulations to predict failure probabilities for different loading scenarios of the pressurized steel tanks, assess the credibility of the predictions, and make a recommendation of whether to retire the tanks. To decide whether the tanks were safe, the probability of failure must be less than 0.1% which indicated a high-fidelity operation. 

Our approach to the Sandia Verification and Validation Challenge Problem was to use probability bounds analysis based on probabilistic representation for aleatory uncertainties (i.e., due to inherent randomness) and interval representation for (most) epistemic uncertainties (i.e., due to lack of knowledge). The nondeterministic model predictions thus take the form of p-boxes, or bounding cumulative distribution functions (CDFs) that contain all possible families of CDFs that could exist within the uncertainty bounds. The scarcity of experimental data provided little support for treatment of all uncertain inputs as purely aleatory uncertainties and also precluded significant calibration of the models. We instead estimated the model form uncertainty at conditions where experimental data were available, and then extrapolated this uncertainty to conditions were no data existed. The Modified Area Validation Metric (MAVM) was employed to estimate the model form uncertainty because the model involved significant simplifications (of both geometric and physical nature). The results of verification and validation processes were treated as additional interval-based uncertainties applied to the nondeterministic model predictions based on which the failure prediction was made.

Relevant publications: 

1. Choudhary A, Voyles I, Roy CJ, Oberkampf WL, Patil M. Probability bounds analysis applied to the Sandia verification and validation challenge problem. Journal of Verification, Validation and Uncertainty Quantification. 2016 February;1(1):011003. doi:10.1115/1.4031285.

Code Verification for the MFIX CFD Code

Status: Complete

Funding agency (Role): US DOE-NETL (while working as a GRA at Virginia Tech)

Summary: In this work, code verification using the Method of Manufactured Solutions (MMS) was performed to verify the main components of an open-source, multiphase CFD code - MFIX. The MMS test cases covered verification of incompressible, steady and unsteady, single-phase and two-phase flows employed within the MFIX suite in an Eulerian-Eulerian framework (called TFM or Two-Fluid Model). The grids used for testing included 2D and 3D, uniform and stretched Cartesian meshes. The no-slip wall, free-slip wall, and pressure outflow boundary conditions were verified. Temporal orders of accuracy for the first-order and second-order time-marching schemes were also assessed. A key challenge in this work was to apply the MMS technique in the presence of divergence-free constraint on the velocity field (for single-phase incompressible model) required by the implementation in this CFD code. Similarly, for two-phase incompressible model, the volume fraction weighted velocity field needed to be divergence free. A novel curl-based manufactured solution was proposed that satisfied the divergence-free constraint during the verification of the single-phase and two-phase incompressible governing equations. Eventually, the MMS test cases were integrated to the continuous integration and testing framework of MFIX. This work contributed heavily to the first-ever Verification and Validation manual of the MFIX software suite.

Relevant publications: 

1. Choudhary A, Roy CJ, Dietiker JF, Shahnam M, Garg R, Musser J. Code verification for multiphase flows using the method of manufactured solutions. International Journal of Multiphase Flow. 2016 April;80(-):150–163. doi:10.1016/j.ijmultiphaseflow.2015.12.006.

2. Musser J, Choudhary A. MFIX Documentation Volume 3: Verification and Validation Manual. Morgantown, West Virginia: US Department of Energy, National Energy Technology Laboratory; 2015.

(Figure) For 3D, steady-state, single-phase flows using curl-based manufactured solution: (a) x-velocity, (b) y-velocity, (c) z-velocity, (d) pressure contours; and observed order of accuracy using (e) L2 and (f) L∞ norms of discretization error. (values in SI units) [Choudhary et al. (2016), International Journal of Multiphase Flow]

Code Verification for Loci/CHEM CFD Code

Status: Complete

Funding agency (Role): NASA-CUIP (while working as a GRA at Virginia Tech)

Summary: To establish confidence in a scientific code, a rigorous assessment of boundary condition implementation is necessary. In this work, techniques were presented for performing code verification of boundary conditions used in a compressible Computational Fluid Dynamics (CFD) code - Loci/CHEM. Loci/CHEM is a finite-volume, compressible CFD code for generalized 3D unstructured grids and employs a density-based solver capable of solving turbulent, multiphase, multispecies, chemically reacting flows, typical of the Aerospace domain (in use at NASA/Marshall Space Flight Center). The goal of this study was to verify the subsonic inflow (isentropic and fixed-mass), subsonic outflow, supersonic outflow, no-slip wall (adiabatic and isothermal), and inviscid slip-wall boundary conditions implemented in the code. The verification exercise used a technique called Method of Manufactured Solutions (MMS) which is a mathematically rigorous way to compare the observed order of the computations to the formal order of the numerical scheme. The use of simplified curved surfaces was proposed for easier generation of manufactured solutions during the verification of certain boundary conditions involving many constraints. To perform rigorous code verification, general grids with mixed cell types at the verified boundary were used. It was found that the use of planar boundaries or only hexahedral cells at the verified boundary can mask sources of errors in the boundary condition implementation.

Relevant publications: 

1. Choudhary A, Roy CJ, Luke EA, Veluri SP. Code verification of boundary conditions for compressible and incompressible computational fluid dynamics codes. Computers & Fluids. 2016 March;126(-):153–169. doi:10.1016/j.compfluid.2015.12.003.

2. Choudhary A, Roy CJ, Luke EA, Veluri SP. Issues in Verifying Boundary Conditions for 3D Unstructured CFD Codes. In: 20th AIAA Computational Fluid Dynamics Conference. Honolulu, Hawaii: American Institute of Aeronautics and Astronautics; 2011. Placed Third in AIAA Student Paper Competition. doi:10.2514/6.2011-3868.

3. Veluri SP, Roy CJ, Choudhary A, Luke EA. Finite Volume Diffusion Operators for Compressible CFD on Unstructured Grids. In: 19th AIAA Computational Fluid Dynamics, Fluid Dynamics and Co-located Conferences. American Institute of Aeronautics and Astronautics; 2009. doi:10.2514/6.2009-4141.

(Figure) Hybrid grids with mixed cell types at the boundaries used for rigorous code verification. (Left) General hybrid grids with sinusoidal boundary surface. (Right) Simplified hybrid grids with spherical cap surfaces to simplify generation of manufactured solutions under many constraints. In some cases, the finest mesh contained ~16.7M (=256*256*256) cells.

Unsteady Simulations of a Flapping Hydrofoil

Status: Complete

Funding Agency (Role): N/A (while working as CFD Engineer at Zeus Numerix Pvt. Ltd.)

Summary: This work involved unsteady CFD simulations of a flapping hydrofoil in viscous, incompressible, turbulent flows. For this purpose, we employed the Moving Frame of Reference (MFR) technique within the RANS solver of Zeus Numerix. The project included designing the solution strategy, modification of solver for moving mesh simulations, generation of grids, execution of simulations, post-processing, analysis, and documentation of results.

Relevant publications: N/A