CFD-experiment validation

Why validation fails quietly

Turbine and channel campaigns often ship a RANS contour plot beside a thermography image and call it validation. The failure mode is subtle: the simulation captures the qualitative trend while missing the quantity that drives maintenance or design decisions, a shifted coolant split, an integrated loss that differs by a few percent, or a Biot-number mismatch that makes wall temperatures agree for the wrong reason.

The HPT leading-edge holing programme at the Oxford Thermofluids Institute pairs engine-scale transient thermography with RANS interpretation of aerodynamic loss and cooling redistribution. That structure, experiment first, simulation as an explanatory layer, not a replacement, is the template this article describes. The editorial home for that research thread is turbomachinery deterioration; the measurement discipline behind it lives under instrumentation & measurement.

A validation workflow that scales

The sequence below is facility-agnostic. Adapt meshing tools and solvers to your stack; keep the documentation structure.

1. Define observables before meshing: list integrated and local quantities the experiment can support (e.g. adiabatic effectiveness area averages, yaw-averaged loss, metal Biot checks)
2. Lock boundary conditions to measurement: inlet total pressure/temperature profiles from rake or probe surveys; wall temperatures from IR or embedded thermocouples; coolant mass flows from metering
3. Run mesh and turbulence sensitivity: document y⁺, domain extent, and turbulence-model spread on a subset of conditions before full matrix submission
4. Compare in nondimensional space: Reynolds, Mach, and Biot numbers from the dimensional analysis module and air properties calculator should match the experiment’s stated similarity case
5. Quantify disagreement: propagate experimental uncertainty (see probe uncertainty note for pneumatic inputs) and report simulation deviation relative to those bands
6. Archive the pair: version-controlled case setup, monitor extracts, and raw experimental reduction referenced from the project or publication record

Turbine hardware: what to match first

For cooled nozzle guide vanes and similar components, prioritise matching coolant-to-mainstream mass-flow ratio and inlet boundary-layer state before chasing local heat-transfer peaks. Leading-edge damage, oxidation holes, eroded edges, changes both aerodynamic loss and coolant redistribution; a pristine-blade simulation cannot validate a deteriorated hardware line without explicit geometry and hole-blockage modelling.

Transient infrared thermography provides time-resolved metal temperature response that RANS can interpret when Biot numbers are handled consistently. The methods note on HPT transient thermography documents data-reduction and uncertainty for that path; the campaign log on the HPT project remains the authoritative record of what was compared and how closely simulation tracked integrated metrics.

Measurement inputs the solver must respect

Pneumatic probe surveys supply inlet profiles and loss coefficients; their bias budgets belong in the validation statement, not in a footnote. Multi-facility calibration, the subject of the 5-hole probe cross-calibration project: exists because facility-specific polynomials will otherwise shift inferred angles and pressures by margins that RANS will happily absorb during tuning.

For a readable treatment of traceability, facility transfer, and similarity, the measurement layer beneath any CFD comparison, see Experimental aerothermal measurement. Articles and notes on lucasrey.com are deliberately cross-linked so a reviewer can move from validation philosophy to derivations to raw project evidence without hitting orphaned pages.

How this fits the knowledge platform

lucasrey.com separates asset types on purpose: peer-reviewed work on publications, campaign evidence on projects, reusable derivations on notes, structured teaching on the fluid mechanics curriculum, and interactive utilities on calculators. Articles like this one connect those layers for human readers, enough narrative to teach a workflow, enough rigour to cite in a methods section, and enough internal linking to strengthen topical authority without keyword stuffing. The HPT deterioration campaign is a concrete example: CFD cases archived on the project page are paired with pneumatic and thermography reductions documented in notes, then synthesised in the aerodynamic and thermal manuscripts listed on the publications hub.