Research

Compressible, reacting and particle-laden flows

I work where high-fidelity simulation meets engineering decisions: writing solvers when the standard ones fall short, verifying them carefully, and using machine learning where it actually saves compute.

Google Scholar Publications

Computational Fluid DynamicsCompressible & Supersonic FlowReacting Flow Simulation (DNS/LES)Multiphase & Particle FlowTurbulence ModellingAeroacousticsMesh-Based Graph Neural NetworksScientific Machine LearningHigh Performance Computing

Ph.D. · McGill · from Jan 2027

Hydrogen–methane flames

Direct numerical and large-eddy simulation of hydrogen-enriched methane flames in the ASEC Lab with Prof. Hendrik Nicolai: how hydrogen blending changes flame structure, stability and turbulence–chemistry interaction. Groundwork so far: a Cantera study from 0D reactors to 1D premixed and non-premixed H₂/air and CH₄/air flames.

DNSLESCombustionCantera

Cantera reactive-flow study

M.A.Sc. thesis · Concordia

Supersonic particle-laden jets & aeroacoustics

Thermal and cold spray accelerate powder through supersonic nozzles at 35–65 bar and 707–807 K. I extended OpenFOAM’s sonicFoam into sonicDPMFoam (two-way coupled Lagrangian particles), ran RANS → DDES on 400+ cores, verified the Kolmogorov cascade, and used the FW-H analogy to link the jet’s acoustic signature to particle distribution.

OpenFOAMDDESDPMFW-HHPC

1 / 12

Featured

Aeroacoustic CFD Simulation of Thermal Spray (Master’s Thesis)

Extended sonicFoam to sonicDPMFoam, an open-source OpenFOAM solver for supersonic flow with two-way coupled Lagrangian particles. Executed RANS → DDES workflows; verified turbulence via Kolmogorov cascade; applied FW-H (libAcoustics) for aeroacoustics.

sonicDPMFoam adds drag, pressure-gradient and virtual-mass forces, wall interaction and optional DEM collisions to sonicFoam; laminar, RANS, LES and DES; serial or MPI. Simulated cold spray process at 35–65 bar and 707–807 K. Developed analytical model for preliminary aeroacoustic analysis. Validated particle–shock interaction and acoustic predictions against literature benchmarks. Processed results with Python/PyVista; managed jobs on Compute Canada HPC (400+ cores).

OpenFOAMDDESAeroacousticsDPM

GitHub arXiv

Scientific machine learning

Graph neural networks as CFD surrogates

Mesh-based GNNs that read an OpenFOAM polyMesh as a graph and predict thermal Homann flow 100× faster than the solver (IEEE Access 2025).

PyTorch GeometricMeshGraphNetOptuna

1 / 10

Featured

Mesh-Based GNN for CFD Acceleration

Achieved 100× speedup over traditional solvers by converting OpenFOAM polyMesh to graph inputs and training Graph Neural Networks for thermal Homann flow prediction. Published in IEEE Access.

Generated large-scale CFD dataset via parameterized CAD, meshing, and OpenFOAM simulation. Converted polyMesh to graph inputs; tuned GNN architectures with Optuna. Achieved 100× speedup over traditional solvers; published in IEEE Access.

PyTorchPyTorch GeometricOpenFOAMOptuna

GitHub Paper

B.Sc. thesis · IUST

CFD-driven design of gas rotameters

Designing and manufacturing a gas rotameter with transient dynamic-mesh CFD in ANSYS Fluent: the float moves with the flow, the mesh deforms with it, and the predicted float position was validated against experiments. Published in Flow Measurement and Instrumentation (2023).

ANSYS FluentDynamic meshTransientValidation

1 / 5

Gas Rotameter Design with CFD

Performed transient ANSYS Fluent simulations with dynamic mesh; validated against experimental data. Published in Flow Measurement and Instrumentation.

Compared triangular vs. quadratic mesh deformation schemes. This was part of my B.Sc. thesis work at Iran University of Science and Technology.

ANSYS FluentDynamic Mesh

Paper

Applied aerodynamics

Supersonic rocket aerodynamics

Aerodynamic coefficients, centre of pressure, nose-tip heating and distributed loads for Space Concordia’s CR25 and two-stage CR26 rockets (Mach 0.8–2.5), validated against RASAero, OpenRocket and flight telemetry.

OpenFOAMcfMeshrhoSimpleFoamValidation

CR25 & CR26 on the Rocketry page →

Publications