The project investigates the acoustic performance of kinked and curved noise screens (Lärmschutzwände, Lsw) that are used to shield residential areas from road‑traffic noise. By combining three‑dimensional finite‑element simulations with analytical modelling, the study aims to provide new calculation modules that can be incorporated into the revised “Guidelines for noise protection at roads – RLS”. The research focuses on how geometric parameters such as kink length and curvature, together with surface impedance variations, influence the insertion loss of a noise screen in the far field on the side opposite the source.

In the simulation part, a realistic line‑source representation of a road was discretised into a series of point sources. The energy contribution of each point source was summed to obtain the total sound field at the receiver location. The three‑dimensional model was then reduced to a 2½‑dimensional analytical description by fitting frequency‑dependent regression curves to the numerical results. The regression analysis was performed for octave‑band centre frequencies ranging from 250 Hz to 4 kHz and for kink lengths of 0.5 m, 1.0 m, 1.5 m and 2.0 m. The coefficient of determination (R²) varied strongly with frequency and kink length. For the smallest kink length of 0.5 m, R² was only 38 % at 500 Hz and 44 % at 250 Hz, indicating a poor fit. In contrast, for kink lengths of 1.5 m and 2.0 m the R² values exceeded 96 % across all octave bands, showing a very good agreement between the regression model and the FEM data. At 1 kHz the regression achieved 97 % for a 1.0 m kink, while at 2 kHz the fit reached 99 % for a 2.0 m kink. For the highest octave band (4 kHz) the regression also yielded R² values above 99 %, although the predicted additional attenuation became slightly negative for very small kink lengths, reflecting a minor over‑prediction of the model in that regime. After filtering the data to exclude outliers and to focus on the most influential input parameters, the overall prediction accuracy improved markedly, with R² values above 92 % for all cases except the 250 Hz band with a 0.5 m kink, where the effect of the kink is negligible because the road‑traffic spectrum is weak at that frequency.

The study also examined the impact of surface impedance. Adding absorbing cladding to the upper edge of the screen consistently increased the insertion loss, especially for the higher octave bands. The results confirm that a combination of curvature, kink length and selective absorption can significantly enhance the acoustic shielding of a noise screen compared with a conventional straight, rigid wall. The research therefore provides a quantitative basis for designing kinked or curved barriers that meet the performance requirements of the updated RLS guidelines.

The project is carried out by a research team in collaboration with the German Federal Office for Transport (BASt). BASt funds the work and provides the regulatory framework that the new models aim to support. The research team is responsible for the numerical simulations, data analysis and the derivation of the analytical expressions. The project spans several years, during which the team iteratively refined the simulation model, performed extensive regression analyses and validated the results against the existing literature on insertion loss and reflection of noise screens. The outcome is a set of analytical formulas that can be directly applied in the 2½‑dimensional calculation method used in the RLS, thereby enabling practitioners to evaluate the acoustic performance of kinked and curved noise barriers with higher accuracy and confidence.