Development of Sonic Crystal Noise Barriers with Simulations and Experimental Validations in Acoustic Chamber

Sonic Crystal Noise Barriers (SCNBs) exploit Bragg scattering in periodic scatterer arrays to attenuate road traffic noise without the visual and aerodynamic penalties of solid walls. Despite growing computational interest, a persistent gap remains between idealised simulation predictions and measured laboratory performance, largely because viscothermal dissipation and diffraction losses are routinely omitted from numerical models. This study presents a systematic, two-phase hybrid methodology to address this gap: first, forty-four Finite Element Method (FEM) configurations were evaluated in COMSOL Multiphysics by varying filling fraction (ff = 0.1-0.5), cylinder diameter (d = 0.02-0.1 m), and lattice constant (α) to identify parameter-dependent trends in A-weighted transmission loss (TLa). Second, fifteen representative prototype configurations were fabricated from acrylic and tested inside an anechoic/reverberation acoustic chamber to validate the computational model across the full 50-4000 Hz range. Simulations predicted a peak TLa of 45.8 dB for ff = 0.3, d = 0.07 m, α = 0.113 m in a 2 × 4 array, while the highest experimentally measured TLa was 10.1 dB for ff = 0.5, d = 0.07 m, α = 0.088 m in a 2 × 6 array. Although quantitative agreement was not achieved, the rank ordering of configurations by TLa was consistent between simulation and experiment across all five filling fractions, confirming the predictive reliability of the computational model for comparative design screening. The observed over-prediction is attributed primarily to viscothermal energy losses, diffraction effects, and material absorption not captured in the idealised 2D lossless model, which reveals that cylinder diameter and lattice constant are the primary drivers of simulated TLa, whereas fill fraction dominates the experimental TLa response, with a 2-row depth identified as the optimal configuration depth for mid-range diameters. These findings provide a validated, data-driven design framework for cost-effective SCNB development and clarify the modelling corrections required to improve quantitative fidelity.