Automated prediction of noise from construction site using stochastic approach

The excessive of noise causes annoyance and suffering to the surrounding neighborhoods. A reliable method of noise prediction is needed to minimize this impact which can be utilized whilst the construction is still in the planning stage and tendering period. This will help the engineers to carry out the proper mitigation method. One such prediction model is stochastic modelling using Monte Carlo approach which has been identified to be able to predict the content of sound for a working day period and has been found to have a good agreement with deterministic method. However, this method is yet to be validated with measurement data and has not been presented in a way that user can utilize them with easy manner. This study investigates the accuracy of the model by comparing the result of the model with measurement of data from real construction sites. Also, this study develops the construction noise prediction tool by using Graphical User Interface, GUI which is useful for the user in noise management. In development of prediction tool, the model was designed to have two parts which are local model and global model that represent the small site and large site respectively. The model has taken into consideration of both the random movement of machines onsite and the random acoustic power of machines for samplings. The model results in the temporal distribution of noise level generated in a working period, including the equivalent noise level, LAeq and the standard deviation. The results were validated with the noise levels for a working day period measured from two different sites which represent a small construction site and a busy construction site. The prediction parameters obtained from measurements on site were applied in current prediction in BS5228 using deterministic approach, for further validation on LAeq. The accuracy of the developed construction noise prediction tool was carried out by means of t-test and mean absolute percentage error (MAPE). It was found that LAeq between measurements, stochastic approach and deterministic were not statistically significant at 0.05. MAPE of temporal distribution noise levels between measurement and modelling showed that the discrepancy was classified in range of 10-50 which was between good and acceptable condition classes. With these results, it showed that stochastic approach can be used as an alternative method of noise prediction to assist the environment engineers as early as in the planning stage in determination of noise arising that may affect the quality of life to the neighborhoods