Raw and processed data for a loudspeaker-coupled thermoacoustic Stirling engine above the Hopf threshold
Description
This dataset supports a study of a thermoacoustic Stirling engine loaded with an audio loudspeaker and operated above its Hopf threshold, where the loudspeaker acts as a linear alternator and the engine sustains a self-excited acoustic limit cycle. It provides the raw measurements, the processed signals, the apparatus specifications and calibration, and the identified model parameters, so that the saturated branch and its identification can be reproduced and independently checked. Data were gathered on a single rig at six supercritical conditions, set by the nominal hot-to-cold temperature difference from 365 K to 415 K in 10 K steps. At each condition the engine was driven from rest to a saturated limit cycle while four channels were sampled synchronously at 3276.8 Hz: the waveguide pressure at x = 0.44 m, the loudspeaker front-end pressure P+ at the front flange (x = -L_d = -1.890 m), the cone velocity from a laser Doppler vibrometer, and the terminal voltage. Both pressures used JTEKT PMS-5M-2 transducers (nominal sensitivity 7.49 Pa/mV); each transducer's own factory calibration was applied, so the records are stored directly in Pa. The heat-exchanger temperature histories are retained as the prescribed thermal forcing. The package has six documented folders (specifications, operating conditions, calibration and identified parameters, computed results, an onset reference, and the raw HDF5 records), with a README and cross-check log. The raw records carry a fundamental near 26 Hz. The fundamental envelope is recovered by demodulating about a fixed carrier and resampling to a 0.5 s grid, and the time axis is reconstructed from the sampling-rate attribute. Every figure and table of the related article reproduces from the processed CSV files without the raw records. Related article: S.-H. Hsu and C.-H. Lai, Adjoint identification of a loudspeaker-coupled thermoacoustic Stirling engine above the Hopf threshold (under review), with an accompanying Data in Brief article.
Files
Steps to reproduce
Apparatus. A thermoacoustic Stirling engine (a looped tube with a stacked-screen regenerator between hot and cold heat exchangers, coupled through a Tee-junction to a branch resonator) is terminated by a Fostex FW168HS moving-coil loudspeaker acting as a linear alternator. The supercritical configuration uses an aluminium-plate piston and no rear housing, with the external electrical load fixed at 500 ohm and 1 microfarad. Instrumentation. Pressure was measured with two JTEKT PMS-5M-2 transducers (nominal 7.49 Pa/mV, each with its own factory calibration), one flush-mounted in the waveguide at x = 0.44 m and one at the loudspeaker front flange at x = -L_d = -1.890 m, conditioned by a JTEKT AA6210 DC amplifier at 2000x gain. Loudspeaker cone velocity was measured with a Polytec PDV-100 laser Doppler vibrometer through the transparent flange; the boundary volume velocity is u' = A_m x velocity, with A_m = 0.01327 m^2. Loudspeaker terminal voltage was recorded. The four dynamic channels were digitised synchronously on an OROS OR35 analyser at 3276.8 Hz. Hot and cold heat-exchanger temperatures were measured with type-K thermocouples and logged on a HIOKI LR8450 recorder. Procedure. Six supercritical conditions were recorded, set by the nominal hot-to-cold temperature difference from 365 K to 415 K in 10 K steps. At each condition the hot heat exchanger was driven so the engine passed from rest, through the Hopf onset, to a saturated limit cycle, giving a 351 s to 657 s transient. The measured temperature histories were retained as the prescribed thermal forcing. Processing. The raw records (fundamental near 26 Hz plus harmonics) are stored directly in physical units (Pa, mm/s, V). The slowly-varying fundamental envelopes of pressure and volume velocity were recovered by demodulating about a fixed carrier near 26.2 Hz and resampling to a 0.5 s grid. The uniform time axis is reconstructed from the sampling-rate attribute, as no time vector is stored. Reproduction. Every figure and table of the related article reproduces from the processed CSV files in folders 01 to 04 without the raw records; the README gives a cross-check log and loading snippets for Python (h5py), MATLAB, and Julia. The model is an operator-based acoustic network (linearised Rott operator, calibrated loudspeaker boundary, Tee-junction) whose parameters are identified by an adjoint method, with the final calibration, Fisher curvature, and leave-one-out cross-validation closed on the genuine nonlinear forward. The identification scripts belong to a separate codebase and are available from the corresponding author on request; ROUTE_C_REPRODUCIBILITY.md documents the exact settings.
Institutions
- National Taipei University of TechnologyTaiwan, Taipei