Annoyance modeling using personal and situational variables for construction site noise in urban areas

Abstract Construction site noise is the most significant cause of noise disputes in South Korea. We conducted an annoyance evaluation to produce annoyance prediction models for four kinds of construction noise that cause the most complaints, namely construction machinery noise from pile drivers, excavators, and concrete pumping vehicles and noise from concrete mold removal working. We adjusted these four noises recorded at construction sites to noise stimuli with 35–80 dB(A), and subjects evaluated annoyance with a score for these stimuli. Participants in the experiment listened to one or two noises at the same time and evaluated their annoyance. We created multiple linear regression models and logistic regression models for annoyance scores of the subjects using acoustic features of noise stimuli and personal and situational variables. Acoustic features included traditional indicators such as Leq, Lmax, and sound quality indices. We calculated mean, maximum, and percentile values of sound quality indices such as loudness, sharpness, roughness, and fluctuation strength. The personal variables were results from a demographic survey and attitudes toward noise, while the situational variables were results from a survey on the living environment and past experiences about noise. Also, we considered the health condition and residence environment of the subjects. Using multiple linear regression, we confirmed that the acoustic features and personal and situational variables influence annoyance from construction noise. Also, we calculated the importance of each variable to the annoyance. Finally, to confirm that the percentage of highly annoyed persons (%HA) varies with personal and situational variables, we classified subjects in accordance with the variables and created a %HA curve for each group. From the result, we confirmed that there is a large difference in %HA in accordance with the health condition of the subjects.