// PUBLISHED RESEARCH

The numerical methods our solvers are built on.

TBL's CFD practice is led by Dr. Abhishek Chintagunta, whose PhD and post-doctoral research at Queen Mary University of London developed the CABARET family of high-order, low-dissipation schemes — the same methods deployed in production when off-the-shelf OpenFOAM and SU2 solvers aren't accurate enough. Eight publications, spanning numerical-scheme design, computational aeroacoustics, compressible multiphase flow, and GPU/HPC acceleration, are summarised below as individual research use cases.

NUMERICAL SCHEME DESIGN COMPUTATIONAL AEROACOUSTICS COMPRESSIBLE & MULTIPHASE FLOW HPC & GPU ACCELERATION

// RESEARCH TIMELINE

Eight publications, one continuous line of work.

Each entry below is a distinct research contribution — read newest first, tracing back to the GPU solver work that first pointed toward high-performance CFD.

2023
NUMERICAL METHODS · TURBOMACHINERY · 14 citations

CABARET on rotating meshes

Extends the CABARET conservative, low-dissipation finite-volume framework to rotating reference frames and dynamic sliding-mesh interfaces — the mesh motion required for turbomachinery, rotorcraft, and wind-turbine simulation. The central result is holding CABARET's low phase-error character across a rotating interface, where conventional ALE and sliding-mesh treatments typically reintroduce the numerical diffusion the scheme was built to avoid. This is the direct methodological link between TBL's published numerical-methods research and its turbomachinery case work — compressor stall margin and turbine secondary-flow loss both sit downstream of this result.

CABARETRotating reference frameSliding meshLow dissipation
2021
NUMERICAL METHODS · COMPRESSIBLE FLOW · 1 citation

Implicit gradients based novel conservative numerical scheme for compressible flows

Introduces an implicit-gradient reconstruction procedure for finite-volume flux evaluation in compressible flow, computing cell-face gradients as part of an implicit solve rather than an explicit least-squares or Green–Gauss pass. The payoff is sharper flux reconstruction near shocks and steep gradients without widening the stencil — the usual cost of higher accuracy — keeping the scheme's computational footprint close to a standard second-order finite-volume method while improving shock-capturing fidelity.

Implicit gradientsFinite volumeShock capturingConservative scheme
2021
NUMERICAL METHODS · MULTIPHASE FLOW

Implicit gradients based novel finite volume scheme for compressible single and multi-component flows

The companion paper extending the implicit-gradient scheme from single-phase compressible flow to multi-component mixtures — flows with more than one fluid species or phase sharing a compressible field, such as multiphase combustion or gas–liquid mixtures under strong compression. Interface sharpness between components is preserved without the spurious pressure oscillations that standard schemes generate at material interfaces under compression, a known failure mode in naive multi-component finite-volume solvers.

Multi-component flowCompressible mixturesInterface capturing
2019
COMPUTATIONAL AEROACOUSTICS · 3 citations

A fourth-order CABARET scheme for Computational Aeroacoustics

Raises CABARET to formal fourth-order spatial accuracy specifically for aeroacoustic wave propagation, where phase error accumulates over the long propagation distances between a noise source and a far-field observer and directly corrupts predicted spectra. The extended compact-stencil operators reach fourth order without sacrificing CABARET's defining low-dissipation, low-dispersion behaviour, validated against canonical linear and nonlinear acoustic wave-propagation benchmarks. This is the scheme variant behind TBL's trailing-edge and jet-noise prediction work.

4th-order accuracyAeroacousticsDispersion errorCABARET
2019
PhD THESIS · QUEEN MARY UNIVERSITY OF LONDON

CABARET Schemes for improved wave propagation and viscous flow modelling

The doctoral thesis unifying the CABARET scheme development programme: the dispersion and dissipation analysis underlying the flux-correction work, its extension to fourth-order accuracy for aeroacoustics, and the treatment of viscous terms needed to move the scheme from pure wave-propagation test cases toward full Navier–Stokes flow. Submitted to the Aerospace Engineering department at Queen Mary University of London, this thesis is the theoretical foundation the subsequent journal papers and the rotating-mesh extension build on directly.

Doctoral researchWave propagationViscous flow modellingQMUL
2018
NUMERICAL METHODS · FLAGSHIP RESULT · 48 citations

Flux-corrected dispersion-improved CABARET schemes for linear and nonlinear wave propagation problems

The most-cited result in the body of work: a flux-correction procedure layered onto the CABARET scheme that suppresses spurious oscillations near discontinuities while retaining the scheme's low-dispersion accuracy on smooth wave propagation. Validated across both linear and nonlinear wave problems, this paper establishes the core scheme variant that every later extension — fourth-order aeroacoustics, rotating meshes, the PhD thesis — builds directly on top of.

Flux correctionDispersion improvementCABARETNonlinear waves
2017
COMPUTATIONAL AEROACOUSTICS · 3 citations

Dispersion improved CABARET for computational aeroacoustics

The first application of dispersion-improved CABARET to aeroacoustic problems specifically, benchmarking phase-error accumulation over long-range sound propagation against standard second-order upwind schemes. This early result is what made the case for a dedicated high-order, low-dissipation scheme in noise prediction — the argument the fourth-order extension and thesis later formalised.

AeroacousticsPhase errorBenchmark study
2014
HIGH-PERFORMANCE COMPUTING

Performance of CFD solver on GPU

The earliest publication in the record, evaluating GPU acceleration of a CFD solver's core computational kernels — flux evaluation and boundary updates, both memory-bandwidth-bound operations well suited to GPU throughput — and quantifying speed-up against multi-core CPU baselines. This is the origin of the HPC and multi-core computing thread that runs through the later work, and the direct precedent for the GPU-accelerated solver runs TBL now offers on large unsteady production cases.

GPU accelerationHPCSolver performance

// FROM RESEARCH TO PRODUCTION

This is why we don't hand you a black box.

Every solver run TBL delivers is traceable back to a published, peer-reviewed method — the CABARET scheme family for high-order wave propagation and aeroacoustics, implicit-gradient finite-volume schemes for compressible and multi-component flow, and GPU/HPC acceleration for turnaround on large unsteady cases. See these methods applied to real engagements in the case studies, or the full publication record on Google Scholar →

// GET IN TOUCH

Have a problem these methods could solve?

Tell us the geometry, the flow regime, and the deadline.