Experimental aerothermal measurement

When experiments fail quietly

In turbine and channel experiments, a few percent bias in inferred total pressure or heat-flux coefficient can reorder competing geometries. The error rarely announces itself as a broken transducer. More often it appears as a smooth trend that “looks physical”, a yaw misalignment absorbed into a polynomial fit, a tap leak smoothed by averaging, or a reference temperature taken at the wrong streamwise station.

The Oxford Thermofluids campaigns documented on this site, from HPT leading-edge holing to multi-facility probe benchmarking: share a common posture: write down the calibration function before the test matrix is frozen, propagate partial derivatives while the dataset is still small enough to inspect, and treat facility transfer as an experiment in its own right. That discipline is what separates a pretty contour plot from evidence that survives peer review and design review.

Traceability from raw signals to published coefficients

Traceability is not bureaucracy. It is the ability to answer, six months later, why row three of the loss table differs from the conference slide. A minimal credible chain looks like this:

1. Raw acquisition: time-stamped voltages or pressures with transducer serial numbers and amplifier settings recorded
2. Calibration artefact: angular or pressure sweeps with fit residuals stored alongside coefficients
3. Reduction: documented equations (not spreadsheet folklore) linking port readings to flow angle, total pressure, Nusselt number, or adiabatic effectiveness
4. Uncertainty: bias and precision separated; sensitivity of outputs to each input, see the 3-hole probe uncertainty note for a worked pneumatic example
5. Archive: version-controlled scripts or notebooks referenced in the publication or project record

On lucasrey.com, short derivations live in technical notes; thematic context is grouped under instrumentation & measurement. Long-form articles like this one connect those fragments into a readable methods narrative for students, collaborators, and reviewers who will never open the raw HDF5 files.

Similarity and nondimensional design

Before scaling a model, decide which Pi groups must match and which can be relaxed with documented justification. Reynolds-number matching governs probe response and boundary-layer state; Mach similarity matters when compressibility couples to heat transfer; Biot and Fourier numbers enter when solid conduction cannot be ignored relative to external convection.

The dimensional analysis module in the fluid mechanics curriculum walks through Buckingham Pi reduction and PDE scaling from first principles, the same logic used when choosing model temperature ratios or deciding whether a room-temperature wind-tunnel result bounds an engine-relevant case. When thermophysical properties enter those estimates, the air properties calculator provides consistent SI outputs for density, viscosity, and Prandtl number across the temperature bands used in gas-turbine and aerothermal rigs.

Facility transfer and calibration ethics

Open-hardware probes and shared calibration rigs make multi-facility studies attractive, and dangerous if transfer is assumed. A polynomial fit valid in Tunnel A can bias Tunnel B when dynamic pressure ratios, plenum unsteadiness, or tap plumbing differ. The correct workflow is pairwise benchmarking on a common reference condition, with disagreement reported as part of the dataset rather than hidden in merged figures.

That is exactly the structure of the 5-hole probe cross-calibration project and the instrumentation thread behind the Oxford open-access probe manuscript. Interactive tooling for bias propagation is planned on the calculators hub; until then, the written derivations remain the authoritative specification.

How this platform documents methods

lucasrey.com is organised for long-term topical authority, not one-off posts. Peer-reviewed work sits on publications; applied campaigns on projects; reusable derivations on notes; structured teaching on the fluid mechanics curriculum; and interactive utilities on calculators. Research themes such as instrumentation & measurement and turbomachinery deterioration provide editorial homes that cross-link those asset types so a reader can move from theory to data to tool without hitting a dead end.

Articles fill the gap between a note and a paper: enough narrative to teach a workflow, enough rigour to cite in a methods section, and enough internal linking to sit inside that graph rather than float as isolated SEO.