A scale-model facility was developed to test the insertion loss (IL) of highway noise barriers. Three model materials were utilized to simulate packed-earth berms and ground (expanded polystyrene), vertical walls (dense polystyrene), and roadways (varnished particleboard). Thirty-eight noise-barrier configurations were tested and used to compare how IL varied with changes to the barrier profile for walls, berms, and combinations of walls and berms for receivers at a representative, highway-adjacent location. The atmospheric conditions were assumed to be homogeneous and nonrefracting. Changes of barrier surface impedance were also assessed. A highway line source was simulated by positioning both an air-jet point source and a receiver microphone at a series of equally spaced points, in order to form an array of source-receiver measurement pairs making differing angles of propagation to the noise-barrier crest line. The IL measurement results are presented in unweighted third-octave bands. In addition, total A-weighted insertion losses (ILA) were obtained by applying an A-weighted, traffic-noise spectrum. When a berm was modeled with surface impedance closely matching that of packed earth, it was found that walls outperformed berms by 1 to 2 dBA. When the surface impedance of a berm was modeled to be acoustically soft, the ILA increased sufficiently to favor berms by about 2 dBA. The result for an acoustically soft berm does not support the long-standing practice of assuming that earth berms outperform walls by 3 dBA, but is consistent with the performance predicted by newer prediction algorithms. When the slopes of berms were made shallower, the IL generally decreased for a berm alone, but generally increased in cases with a wall atop the berm.

Besides numerical models, another way to model complex environments is using physical reduced-scale models. This project used scale models to study several factors which may influence the performance of roadside noise barriers. One is the barrier absorption. Making the barrier sound absorptive decreases reflections and may decrease amplification between parallel barriers. The other is tree foliage growing near the barrier. Tree foliage may scatter sound into the shadow zone behind the barrier, increasing noise levels and decreasing the insertion loss. Tree foliage may also attenuate sound that would normally be diffracted into the shadow zone, actually increasing the insertion loss. A 1:31.5 scale model was created in an anechoic chamber to test the effects of these two factors. Excess attenuation measurements were performed to choose scale model materials, which accurately represent full scale surfaces. Parallel barriers were modeled with a reflective surface and their IL's were measured for different configurations of absorptive covering. IL's of single barriers were measured both with and without scale model trees placed either in front or behind them. The results are compared to previously performed field test measurements.

France 4097This paper describes a field experimental investigation of the effects of nearby vegetation on the performance of roadside noise barriers. The effects of nearby vegetation (trees, hedges, vines, etc.) are unclear. It may decrease noise levels behind a barrier, either by back-scattering or absorbing sound. Foliage may also increase noise levels by scattering sound which would normally pass above the barrier into the shadow zone. Field test sites were identified to study these effects. Traffic noise levels were measured simultaneously behind foliage and no-foliage cases and the results were compared. Both increased and decreased noise levels behind the barrier were seen in different frequency ranges, with most effects being less than 5 dB.

No abstract

Statistical energy analysis (SEA) has been recently used to predict the sound transmission loss of single and double partitions [Price and Crocker, J. Acoust. Soc. Am. 47, 683 (1970)]. The present authors have applied a similar SEA technique to the problem of sound transmission between two adjacent rooms whose walls, floor, and ceiling are all structurally coupled. The model examined consisted of a rectangular box separated into two cavities by a dividing partition. The SEA model therefore consisted of five coupled resonant elements: the source room, the source and receiving room common sidewalls (assumed to be identical in each of the two rooms), the dividing partition, and the receiving room. The theory considers the effects of the material thickness and damping of both the dividing partition and the coupled sidewalls. Wave interaction at the structural junctions of the partition and the sidewalls was also examined and used to predict the required structural coupling loss coefficients. The theory was found to produce good agreement from 500 to 16 000 Hz (±2 dB) with experimental data obtained from a pair of model rooms. It is shown that by careful selection of system parameters, the effects of structure-borne flanking can be minimized.