Low-frequency impact sound in CLT floors: Perceptual gains from beam integration

Abstract

Low-frequency floor-impact noise is the principal acoustic shortcoming of cross-laminated-timber (CLT) construction, yet current single-number ratings are rooted in tapping-machine spectra and under-represent heavy/soft human impacts below 100 Hz. We couple a validated plate–beam finite-element model of a two-story CLT test building to a rectangular room-acoustic model and generate 12 auralized stimuli—three slab thicknesses (150, 210, and 270 mm) with and without mid-span and quarter-span beams, each excited by literature-based jump-type (1.2 kN, 20 ms) and run-type (0.6 kN, 30 ms) forces. The frequency-domain responses are converted to time waveforms via inverse FFT and reproduced, after full transfer-path equalization, to 20 normal-hearing adults in a semi-anechoic room. Psychophysical scaling employs magnitude estimation (ME) for all stimuli and paired comparison (PC) for the six jump-type cases, while the objective descriptor is the event maximum level L_AFmax (ISO 717-2:2020, Annex D). Beam integration lowers L_AFmax by 1.9, 1.3, and 5.9 dB for the 150, 210, and 270 mm slabs, respectively, with a mean reduction of 3.1 dB; thicker slabs are consistently quieter than thinner ones. ME geometric means correlate strongly with L_AFmax (r = 0.88, semi-log fit, R² = 0.77), and the PC loudness scale shows a comparable association (r = 0.85, R² = 0.73), confirming that event-level attenuation translates directly to perceived quietness. Results demonstrate that beam integration is an effective structural lever for sub-100 Hz impact noise and that L_AFmax, evaluated on simulation or rubber-ball tests, provides a perceptually grounded design metric that complements conventional ratings.