I develop stochastic, Lagrangian, and computational frameworks to understand why turbulent flows generate sound — and how to make them quieter. Physics-first, always.
How does the chaotic, multiscale structure of turbulence produce sound — and can we predict, model, and control it without solving everything at once?
My research sits at the intersection of fluid mechanics, mathematical physics, and numerical methods. The approach spans three tracks: mathematical modelling (stochastic Markov-chain turbulence, vortex particle methods, Lighthill's and Lattice Boltzmann analogies), computational simulation (LES, DES, LBM, FW-H), and physical experiments (Acoustic Wind Tunnel Braunschweig, porous coatings, airfoil–turbulence interaction).
The result is a programme that can move between a tensor identity and a wind turbine blade without losing rigour — or between a Markov chain and a microphone array without losing contact with experiment.
Stochastic vortex particle methods for predicting broadband airfoil noise from incoming turbulence. Time-domain BEM validated against AWB experiments.
Explore →One-dimensional turbulence as reduced-order model for jet self-noise. Markov-chain representations coupled to Lighthill's analogy with quantifiable uncertainty.
Explore →How porous coatings and cross-sectional geometry control vortex shedding noise. Permeability, thickness, and shape as aeroacoustic design parameters.
Explore →Turbulence distortion and coherence effects in airfoil leading-edge noise via LBM. New computational frontier at DLR for mesoscopic aeroacoustics.
Explore →Self-similarity, far-downstream velocity statistics, and acoustic emission from turbulent round jets at high Reynolds number using ODT and LES.
Explore →Turbulent inflow noise from wind turbine blades. AWB experiments, multi-scale LES, and noise prediction within DLR's HGF energy programme.
Explore →Full annotated list → Publications · Google Scholar · ResearchGate
Physics-based mathematical and computational frameworks for understanding, predicting, and controlling the noise generated by turbulent flows. The problems are fundamental; the tools are often novel — stochastic Lagrangian methods, mesoscopic Lattice Boltzmann, one-dimensional turbulence, and high-fidelity LES, always anchored in experiment.
Markov chains, Langevin equations, Monte Carlo, ODT for turbulence statistics without resolving all scales.
Lagrangian meshless Euler/N-S solvers. Lattice Boltzmann for mesoscopic turbulence–sound coupling.
LES, DES, DNS in OpenFOAM and SU2. Scale-resolving for aeroacoustic source analysis and model validation.
Lighthill, FW-H, BEM. Connecting flow solutions to far-field radiation in a mathematically rigorous way.
This thread runs from my doctoral thesis through the most recent DLR work. I developed a stochastic, time-domain approach using vortex particles as both a turbulence representation and an acoustic source model. A Lagrangian description of turbulence — individual vortex elements with prescribed statistics — is physically transparent and computationally tractable.
The framework produces broadband noise predictions comparing well with high-fidelity LES at orders-of-magnitude lower cost, validated against AWB wind-tunnel measurements. A look-up table extension (DLR, 2024) scales this to operational aircraft and wind turbine applications.
The most recent work (WTN 2025, AIAA/CEAS 2026 LBM paper) explores turbulence distortion and coherence effects through airfoil interaction using Lattice Boltzmann simulations — a new frontier where mesoscopic physics governs the broadband noise spectrum.
One-Dimensional Turbulence (ODT) directly simulates turbulent advection via stochastic rearrangement events on a 1D domain, preserving the physics of the energy cascade without 3D DNS cost. My DFG Walter Benjamin work at Cambridge applied ODT to turbulent jet dynamics: self-similar velocity statistics, far-downstream fluctuations, and acoustic emission via Lighthill's equation.
Key result (GAMM 2023): ODT-resolved acoustic sources at high Reynolds number — bridging the gap between low-order turbulence models and the acoustic field with explicit probabilistic treatment of source fluctuations and quantifiable uncertainty.
Porous coatings offer a passive, structurally simple route to noise control. Owls fly nearly silently using porous wing structures. The engineering challenge is to understand the mechanism precisely enough to design effective coatings without trial and error.
My computational studies (DES, FW-H) on porous-coated cylinders show permeability and thickness act on distinct mechanisms. The optimal combination is Reynolds-number dependent. Recent work extends to non-circular cross-sections: trapezoidal cylinders (AIAA/CEAS 2023), and now square cylinders with porous coatings — both experimental and numerical investigations presented at AIAA/CEAS 2026. Applicable to airframe landing-gear and wind turbine strut noise.
Complementing the modelling track: LES/DES in OpenFOAM and SU2, FW-H post-processing, and grid-generated turbulence simulations benchmarked against AWB wind-tunnel data. A dedicated study (STAB/DGLR 2024/2026) characterises airflow turbulence in the AWB using turbulence grids — providing the experimental ground truth against which all simulation approaches are validated.
Organized by year, annotated for significance. Live profiles: Google Scholar · ResearchGate · ORCID 0009-0000-1911-9317
Generates turbulent inflow conditions using Gaussian-profile vortex elements. Produces statistically correct energy spectra for leading-edge noise simulations.
2D meshless Lagrangian solver for incompressible Euler equations. Core tool for the time-domain leading-edge noise framework.
Computes directivity patterns of radiated sound pressure from airfoils using the Ffowcs Williams–Hawkings acoustic analogy.
I believe computational results should be reproducible. Where journal policies and data agreements permit, code and data associated with published papers are made available. If you are trying to reproduce a specific result and cannot find the code, get in touch.
I am a research scientist at the Institute of Aerodynamics and Flow Technology at DLR Braunschweig, working in aeroacoustics on turbulent noise from aircraft high-lift systems and wind turbines.
My research is unified by a single question: how does the chaotic, multi-scale structure of turbulence give rise to sound — and can we model this mathematically, without losing physical meaning? This has led me to develop Lagrangian vortex particle methods, stochastic Markov-chain turbulence models, one-dimensional turbulence (ODT) approaches, and most recently Lattice Boltzmann simulations — each chosen for what it reveals about the turbulence–sound coupling, not for computational fashion.
Before DLR, I held a DFG Walter Benjamin Postdoctoral Fellowship at the Department of Applied Mathematics and Theoretical Physics at the University of Cambridge (2021–2023), working on the mathematics of jet noise with Dr. Lorna Ayton and Prof. Marten Klein. The Cambridge period deepened the connection between mathematical physics and turbulence modelling — particularly through ODT and Lighthill's acoustic analogy.
My PhD (summa cum laude, BTU Cottbus & TU Berlin, 2016–2019) established the time-domain stochastic vortex particle method for predicting leading-edge noise — recognised with the BTU Best Thesis Award 2020 and DEGA Young Scientist Award 2019. I worked at Rolls-Royce Aeroengines UTC on applied aeroacoustics between academic positions.
Computational Fluid Dynamics (advanced postgraduate lecture course), BTU Cottbus — winter terms 2020/21, 2021/22, 2022/23. Workshop on CFD and Aeroacoustic Analogies (with Dr. T. Geyer), December 2020. Teaching assistant, Engineering Mathematics, BTU Cottbus, 2020–2021.
Based at DLR Braunschweig. Always open to scientific discussion, collaboration proposals, or speaking invitations.
Interested in a master's thesis or internship in computational aeroacoustics or turbulence modelling? Send a brief note on your background and interests with your CV. Strong mathematical background and Python or C++ experience are particularly welcome.
Real-time numerical visualisations of the turbulence physics underlying my research — built directly from the governing equations. Use the controls to explore the parameter space.
A vortex tube in a turbulent background field — progressing through the complete lifecycle of a coherent structure in HIT: initial ring, strain induction, vortex stretching, Kelvin-Helmholtz kinking, reconnection, and Kolmogorov-scale decay.
The vortex stretching term ω·∇u amplifies vorticity along principal strain axes — thinning the tube, growing enstrophy, and driving the energy cascade to smaller scales.
Vorticity preferentially aligns with the intermediate strain eigenvector in HIT (Ashurst et al. 1987) — a statistical signature reproduced here.
28 Fourier modes with random phases generate the HIT velocity field — the same synthetic turbulence approach used in my vortex particle leading-edge noise model.
This lifecycle — from coherent ring through stretching and reconnection to dissipation — is the physical process that determines the turbulent inflow statistics my stochastic vortex particle methods must capture to predict leading-edge noise correctly. The Fourier-mode synthetic turbulence shown here is the backbone of the method in Sharma & Sarradj, PRF 2019 ↗
ODT jet turbulence · LBM turbulence–airfoil interaction · Porous cylinder vortex shedding
Intuitive, rigorous guides to the mathematical frameworks underlying my research — written for physicists, mathematicians, and engineers who want more than a textbook summary.
Rapid Distortion Theory — the linearised turbulence framework that underlies Amiet-type leading-edge noise models and sets the validity boundary for LBM simulations.
When turbulent air is squeezed and stretched as it flows around an object, its swirls change shape. Rapid distortion theory is the surprisingly simple idea that, if this happens fast enough, you can predict exactly how — with nothing more than straight-line maths.
Picture turbulent air not as chaos, but as a busy crowd of small spinning swirls — eddies — of every size, all jumbled together. On their own each one quietly spins and slowly fades.
Now push that crowd through a region where the flow is being strained — stretched in one direction and squeezed in another, like air funnelling into a narrowing duct, or sweeping around the nose of a wing. Every eddy gets deformed by that straining. A round swirl is pulled into an oval. And here is the part that matters: as a swirl is stretched thin, it has to spin faster — the same reason a spinning skater speeds up by pulling their arms in.
Drag the slider below and watch a field of round, evenly-mixed eddies get distorted by a steady strain.
Notice what the strain did: it took turbulence that looked the same in every direction and made it directional — stronger across the flow than along it. That directionality is called anisotropy, and it is the single most important thing RDT predicts. The theory works out this stretching for one swirl size at a time, then simply adds all the sizes back together.
The whole theory rests on one bet: that the distortion happens fast. There are two competing clocks.
Clock A — the distortion time. How long the air spends being strained as it passes through the squeeze.
Clock B — the eddy turn-over time. How long an eddy takes to interact with its neighbours, trade energy, and reorganise itself.
If Clock A is much shorter than Clock B — the distortion is over before the eddies have time to "talk" to each other — then each eddy is just passively stretched, and the simple linear maths is exact. This is the rapid in rapid distortion theory. If the distortion is slow, the eddies do interact while being strained, and the theory only gives a rough guess.
Choose how fast the air is pushed through the straining region:
RDT was not invented in one go. It grew from wind-tunnel engineering — the practical question of why turbulence calms down when it is funnelled through a contraction.
Works out how a single wave-like swirl is changed by stretching a stream — the seed of the whole method. [5]
Add up Taylor's result over every swirl size to give the first full theory of how rapid straining distorts real turbulence. The method gets its name. [1]
Extends RDT from a uniform squeeze to the flow around a body — a cylinder, and by extension a wing nose. This is the version aeroacoustics still builds on. [2]
Set out clearly when RDT can be trusted and when it cannot — the "two clocks" rule made rigorous. [3]
Why does an aeroacoustician care? Because when turbulent air strikes the leading edge of a wing or blade, it generates broadband noise — a major source of the sound from fans, turbines and aircraft. The standard way to predict that noise (Amiet's theory [4]) takes the incoming turbulence as its input and assumes it arrives unchanged.
But it does not arrive unchanged. As we saw in §1, the wing's nose strains the eddies before they ever reach the edge — squeezing, stretching, redistributing their energy. RDT is the tool that predicts this pre-impact reshaping. Feeding the distorted turbulence into the noise model, rather than the raw upstream turbulence, measurably improves predictions [6][7].
RDT in this setting is usually used only in its two easy extremes — very small eddies, or very large ones — and those who introduced it were careful to say the correction could not be assumed general [6]. Closer to the edge, where the straining is strong and prolonged, even the RDT-based prediction can drift from measurement because it does not fully capture the distorted turbulence [7]. Sharpening that account — explaining, not just curve-fitting, what the leading edge does to turbulence — is exactly the open problem this primer's author works on.
Every reference below has been checked against the original publication record. The intuition in this primer is owed entirely to the authors named here.
Foundational precursors not separately listed: Prandtl's contraction analysis and Ribner & Tucker's (1953) shock–turbulence work, both cited within references [1] and [2].
Lighthill’s acoustic analogy · Sears function & gust response · Stochastic turbulence & Itô calculus · ODT — the mathematics