Resolved numerical simulations
Verified interface, wetting, transport, and conservation methods for unstructured finite-volume calculations.
Agentic computational science · verified engineering
Verified agentic discovery and design for thermofluid systems.
I help industrial R&D teams improve heat-transfer and multiphase-flow systems—with quantified uncertainty. Agentic workflows accelerate CFD model development, testing, and optimization; verification, provenance, and human review show which improvements can be trusted.
Industrial microcavity fillingexperiment × 3D VOF
23peer-reviewed contributions
5fully industry-funded doctoral projects
4connected Bosch microfluidics studies
6completed doctorates as primary supervisor
Primary design service
Topology, shape, and parameter optimization for heat-transfer and multiphase-flow systems.
Agentic workflows coordinate model development, solver execution, verification, evidence extraction, and auditable design records. Bayesian optimization directs expensive evaluations. Open-source CFD resolves the governing physics. Feasibility, convergence, conservation, provenance, and engineering judgment remain visible in every accepted result.
- Application range
- Conjugate heat transfer, thermal management, microfluidic filling, wetting, mixing, and multiphase transport.
- Evidence today
- OpenFOAM work since 2007, co-authorship of The OpenFOAM Technology Primer, a CAD-to-OpenFOAM CHT evaluator, verified multiphase methods, and industrial microfluidic studies.
- Current stage
- MVP development: expert-led design loops with reusable evaluators, acceptance gates, provenance records, and agent orchestration.
- Best first step
- A design brief describing the geometry, operating envelope, objective, constraints, available evidence, and whether an NDA is required before technical details are shared.
Connected capabilities
From physical fidelity to operational confidence.
Each layer strengthens the next: resolved physics generates meaningful data; probabilistic models expose uncertainty; optimization converts both into action.
Complex industrial multiphase flows
Microfluidic filling, viscoelastic interfaces, surfactants, mass transfer, and capillary flows in consequential geometries.
Experimental and production evidence
High-speed imaging, automated measurements, and process data that reveal missing physics and operating variation.
Probabilistic learning and Bayesian optimization
Uncertainty-aware surrogates and gated optimization when simulations and experiments are expensive.
Selected results
Methods designed for the edge cases that matter.
Wetting, pinning, stable simulation, and filling regimes form one evidence chain.
High-speed experiments, automated image analysis, and resolved geometric-VOF simulations establish the foundations for industrial microfluidic digital twins.
Explore the four-study programMass and momentum must cross an interface consistently.
Peer-reviewed LENT and geometric-VOF formulations remain stable where small discrete inconsistencies can otherwise dominate the result.
Explore the numerical evidenceSelected publications
Publications connected to data, source code, and reproducible evidence.
The complete record spans multiphase methods, microfluidic experiments, scalable CFD + ML, and research infrastructure—with direct links to the supporting research outputs wherever they are publicly available.
A residual-based non-orthogonality correction for force-balanced unstructured volume-of-fluid methods
A deterministic residual-based non-orthogonality correction restores surface-tension and gravity force balance near linear-solver tolerance on polyhedral meshes.
Combining machine learning with computational fluid dynamics using OpenFOAM and SmartSim
A scalable, in-memory coupling enables online training and inference between parallel OpenFOAM simulations and machine-learning models.
Experimental and numerical study of microcavity filling regimes for Lab-on-a-Chip applications
High-speed experiments and three-dimensional geometric-VOF simulations identify distinct microcavity-filling regimes and the conditions controlling gas entrapment.
Teaching
Mathematics, numerical methods, and scientific computing.
Courses and open materials connect mathematical foundations with reproducible engineering practice—from Mathematics II to Machine Learning in Fluid Mechanics and OpenFOAM–ML hackathons.
Teaching & supervision →Computational practice
Research software and agentic workflows as shared infrastructure.
OpenFOAM Data-Driven Modelling, LEIA/LENT, DeboRheo, SFB 1194, NFDI4Ing, and current agentic workflows connect methods, people, and reusable evidence. Agents help develop, test, document, and audit; verification and human judgment decide what is accepted.
Explore the workflow →A hard thermofluid problem?