Noise prediction for infrastructure construction activities using simple prediction chart technique

Construction noise is one of the most severe pollutions in the environment and causes different physical and psychological damages to human beings. In recent years, the research archetype has transposed towards the prediction of construction noise for its adverse effects on the environment and construction workers. Hence, a reliable noise prediction method is mandatory to control, mitigate, and abate the potential hazardous noise. This study aims to validate the results between the predicted and actual noise levels of construction activities mainly on infrastructure works by using the simple prediction chart technique. This study was carried out by using the sound power levels emitted by an individual machine during different infrastructure construction activities to predict the noise levels at different locations. The results showed a high accuracy of predicted noise levels along with an absolute difference of less than 3.0 dBA and a relative error of less than 4.0 %. Besides, the predicted noise levels are reliable as the R-squared value was high. On that account, the simple prediction chart method technique has the potential to be utilized as a managerial tool that may help to reduce the negative impacts of environmental noise to the surroundings and occupational noise to the workers.


Introduction
Noise-induced by construction activities is claimed to be one of the major noise pollutions in both developed and developing countries [1][2][3]. Many studies proved that construction activities generate high levels of fluctuating and continuous noise that may cause disturbance to sensitive areas [4][5][6][7][8]. Although every construction stage has contributed to noise pollution at different levels, substructure works are proved to be the noisiest stage among all [8,9].
In the United States, an estimation of 22 million workers is currently exposed to perilous occupational noise [10]. Moreover, 11.2 million workers in Canada are overexposed to occupational noise and one-third of the workers experienced acute effects of hearing loss [11]. A study estimated prolonged occupational noise exposure accounts for 16 % of hearing impairment in adults globally [12]. Besides, construction workers at the workplace were exposed to noise levels exceeding the permissible noise level of 85 dBA [13]. Johnson and Morata [14] mentioned that 18 to 22 % of construction workers were under the exposure of occupational noise above 85 dBA during working hours. In addition, a quantitative assessment revealed that the majority of the heavy machine operators are exposed to noise levels exceeding 85 dBA [15]. Heavy machines that generate noise levels exceeding 85 dBA may cause risks of overexposure among workers and operators. Gan and Mannino [16] proved that the prevalence of bilateral high-frequency hearing loss is correlated to cumulative occupational noise exposure time, which explained that workers that were exposed to a high level of noise for a long duration, are more likely to suffer from hearing impairment. Hence, irreversible noise-induced hearing loss (NIHL) problems among construction workers often arise due to overexposure to loud noise [17].
A study proclaimed that construction sites are one of the sources that create noise pollution in the surroundings [3]. Foo [2] claimed that the construction industry was identified as the second most impactful source that contributed to noise pollution in the surroundings. The main sources of noise at a construction site consists of construction machines that emit high levels of noise such as earth-moving machines and pneumatically driven devices [18]. With regards to the impact of occupational noise and environmental noise as discussed, many researchers developed different noise prediction methods to predict excessive noise from construction activities [4,7,[18][19][20]. Construction noise prediction is usually conducted to monitor, control and assess the noise source in a construction site. The outcome of the studies proved that noise prediction models are reliable and accurate as compared to the deterministic approach such as BS 5228 -Part 1: 2009. Hence, these noise prediction models can be utilized as a managerial and planning tool to prevent and mitigate potential noise hazards from the workplace. It is the obligation of management to ensure the noise level is maintained under an acceptable limit in the workplace but the importance of noise management was neglected and resulted from ineffective noise attenuation at construction sites [21].
This study introduced the application of the simple prediction chart method to predict the noise levels of different infrastructure construction activities. The simple prediction charts technique mainly considered the random movement of the machines that operate at full power and the distance between noise source and receiver. Six case studies were carried out for data validation by assessing the difference between the prediction and actual measurement. Substantially, this paper revealed the accuracy and reliability of the simple prediction technique.

Application of simple prediction chart method
Construction noise prediction is very essential in the planning phase of construction to ensure the potential noise hazards can be eliminated and mitigation of noise can be made [22]. The prediction included several factors such as sound power of the machine, the aspect ratio of the site, the distance between receiver and noise source, variation of angle from the site centre, full power operation of the machine and the reduction due to noise absorption of the earth [7]. Haron et al. [7] developed the simple prediction chart method which can determine the noise levels at several different locations both manually and quickly, by adopting the sound power level of different machines that may be operated in actual construction activity. Findings showed that the simple prediction chart technique is capable of providing reasonable and reliable outcomes and has the absolute differences of 3 dBA [7]. The most crucial factor when applying the simple prediction chart method is to identify the overall size of the construction site and divide the site into various sub-areas, then the noise levels of the respective areas will be predicted and combined to obtain the overall noise levels of the construction site. Besides, multiple machines that are being operated for different activities can be clustered within a sub-area. The sound pressure levels at different points in the construction site are predicted using the simple prediction chart technique. However, the position of all the control points was selected based on the accessibility of the location. The determination of the site aspect ratio of the case studies was determined based on the movement area of the earth-moving machine.
The sound pressure level at a receiver can be obtained by using the following seven steps: (1) select the sound power level of the machine; (2) determine the width and depth of the sub-area; (3) identify the angle away from the sub-area centre; (4) compute the distance between the plant and receiver and the r/w ratio; (5) determine the standard deviation by referring the simple prediction chart [7]; (6) determine the mean level deviation by referring the simple prediction charts [7]; (7) calculate the mean noise level by using equation (1). Lastly, combine the mean noise levels from each sub-area by applying equation (2) to obtain the equivalent mean noise levels and equation (3) to compute the combined standard deviation. L = Lw -20log10(r)-8+ΔL ( 1 ) Where L = mean level, sound pressure level corresponds to the source at centre of site (dBA), L w = sound power level (dBA), r = distance between receiver and center of sub-area (m), ∆L = mean level deviation (dBA). L AFeqn = 10.log 10 (10 Lp1/10 +10 L2/10 + 10 Lpn/10 ) ( 2 ) Where L p1 , L p2 ,…L pn is the mean noise level of each machine calculated by using Equation 1.
Where σ 1 , σ 2 ,…, σ n , is the standard deviation of the mean noise level for each machine.

Methodology of site measurement
This application of the simple prediction charts technique mainly focused on the infrastructure construction activities. A total number of six case studies with different parameters such as the site aspect ratio, the coverage angle, the distance between the noise receiver and the site centre, and the earth-moving machine duty cycles were carried out at the residential projects in Semenyih, Selangor, Malaysia. All the noise level measurement procedures are in accordance with BS ISO 6395:2008, BS 5228-1:2009, and BS ISO 3744:2010 [23][24][25]. The noise indices considered were the equivalent continuous Aweighted sound pressure level (L AFeq ), maximum sound pressure level (L max ), minimum sound pressure level (L min ), sound pressure level exceeding 10 %, 50 % and 90 % of the time of measurement duration (L 10 , L 50 , L 90 ).
A Type 1 sound level meter SoundTrack LxT of Larson Davis was used to conduct all the noise level measurements. The sound calibrator with the reference sound of 94 dBA at 1 kHz was used to calibrate the sound level meter with a difference of 1 dBA before conducting the measurement. The sound level meter was set at 1.2 m to 1.5 m away from the ground level and 3.5 m away from the reflective structure as stated in Guidelines for Environmental Noise Limits and Control [26]. Background noise and the value of equivalent continuous A-weighted sound pressure level, L AFeq of each control point was measured for 30 minutes using the short sampling method [26]. To conduct the measurement of the sound power level of the machine, the basic length, one of the machines will be measured and the radius, r will be determined according to BS ISO 6395:2008 Annex A [23][24][25]. Each location of microphones was calculated by using a set of coordinate systems. The sound power level of the machine was obtained at 6 locations with different coordinates around the machine. The measurement duration at each point was 30 seconds as proposed by previous researchers [18][19][20].
The absolute difference value between prediction and actual measurement was used to determine the accuracy of the prediction results as this method was adopted in data validation by the previous researchers [19]. The R-squared method was applied in this study as this technique was used by previous research in determining the reliability of noise prediction results [27]. The strength of association of the R-squared value consists of 5 categories of strength which are very strong, moderate to strong, weak to moderate, and weak to comprehend the association of the data [28].

Noise emission levels from individual machine
The determination of sound power level (L w ) of the machines for earth-moving machinery was derived from British Standards BS ISO 6395: 2008. The sound power levels were used as an input to predict the equivalent continuous A-weighted sound pressure level (L AFeq ). The important information of the machines such as the machine manufacturer, machine model number, engine net power at corresponding revolution per minute was recorded and presented in Table 1 as stated in BS ISO 6395:2008 and the information was verified by the personnel of the construction sites. Table 2 tabulated the input data of sound power level and the results of calculation for the sound power level of each machine. All the machine emitted noise levels were recorded during the lunch break so that the irrelevant noise sources were excluded during the measurement. The machine emitted noise levels were applied to calculate the L p ' (ST) of the machines. Next, ΔL p was computed by finding the difference between the L p ' (ST) and the mean equivalent background noise, L p(B) . The ΔL p of all the machines were greater than 15 dBA, which indicated that the loudness of the background noise was insignificant to cause an impact on the determination of sound power level. Consequently, the background noise correction factor, K 1 for all the machines was determined as 0.

Background noise of the construction sites
The background noise of each construction site was recorded before the commencement of any construction activities; the measurement was conducted at 7.00 a.m. and lasted for 30 minutes. Case studies 6 had the highest background noise of 54.1 dBA because there were existing buildings located 30 m away from the sub-area that may reflect the noise. Next, case study 1 had the background noise of 54.0 dBA due to the traffic noise as the sub-area was located 20 m away from the entrance of the construction site. Next, the background noise for case study 5 was 51.1 dBA and that resulted from the loud natural ambient noise of the animals and birds at the same construction site. For case study 4, the background noise is slightly lower than the previous case studies because the construction activity was located in a residential area. Lastly, case studies 2 and 3 had the lowest background noise of 47.0 dBA because the sub-area was an isolated area located 50 m away from other construction activities and the background noise of both case studies was taken at the same construction site.

Noise emission levels from infrastructure construction activities
Based on the results, case study 4 had the highest noise levels of 79.8 dBA and a maximum sound level of 105.2 dBA. This phenomenon resulted from high impact noise caused by the excavator during the sheet pile installation. The operators were using earmuffs whereas the other four workers which were the supervisor, pipe welder, traffic controller and the general worker did not equip themselves with hearing protection equipment throughout the working duration of 8 hours. Prolong exposure to high noise levels may result in workers suffering from noise-induced hearing loss problems [14,16,29].
The lowest noise levels among all the case studies were found to be 69.3 dBA for case study 5. Although three earth-moving machines such as crawler excavator, road roller and back pusher were operated in the road construction activity, there was no high impact noise being generated during the activity. The noise was mainly induced from soil compaction and placement of crusher run because only minimal excavation works were conducted. There were three operators equipped with hearing protection devices and general workers were not involved in this activity. The noise levels of case studies 1, 2, 3 and 6 range between 69.3 dBA and 77.8 dBA, and each case study consisted of a machine operator and two to four workers. However, hearing protection devices were absent among the operators and workers. This phenomenon arose from a lack of personal safety concerns [10], and the hearing protection devices were not provided to the construction workers.

Construction noise prediction using simple prediction charts methodology of site measurement
As shown in Table 3, the highest noise levels of 78.4 dBA were induced by the combination of two crawler excavators during the water distribution system construction activity in CS 4. In contrast, CS 3 had the lowest sound pressure levels of 70.3 dBA. Figure  1 demonstrates the illustration of the site layout of case study 1. The measurement for control point 2 in case study 1 is depicted in Figure 2.
For case study 1, the location of control points 1, 2 and 3 were selected because there were houses located on the right side of the sub-area as depicted in Figure 2, and the selected locations were at least 3.5 m away from reflective structures and 30 m away from other noise sources. The selection of control points 1, 2 and 3 for case studies 2 and 3 was to assess the noise levels when the control points are located parallelly to the sub-area with different distances. For case study 4, control points 1 and 2 were located close to the centre of the site whereas control point 3 was placed slightly farther from the site centre. The configuration was intended to avoid the interference of vehicles because the sub-area was located right next to an access road. The control points of case studies 5 and 6 were chosen to cover the movement path of the earth-moving machines throughout the observation duration.     Figure 3 and Table 4 show the disparities of predicted noise level and actual noise level of every control point for all the case studies. Besides, the accuracy of the predicted results is tested by using the absolute difference and relative error. The disparities of the comparison have the highest absolute difference of 2.9 dBA and the lowest value was 1.0 dBA. The highest relative error among the case studies was 4.0 % and the lowest relative error was 0.8 %. Overall, case study 4 contributed the lowest absolute difference among the results whereas the absolute difference of case study 5 was the highest among all the case studies. By interpreting the data, the simple prediction chart has high accuracy from each case studies with the mean absolute difference ranging from 1.2 dBA to 2.0 dBA. The disparities between the prediction and actual measurement of case studies 2, 3 and 5 can be explained by the coverage area of the machine during the activities was insufficient because this technique assumed the machines to cover all the areas within the well-defined sub-area [7]. Besides, a study revealed that the accuracy of the prediction is directly affected by the coverage area of moving machines. 19 For case studies 1, 4 and 6 the machines were moving to almost every area within the sub-site and hence resulted in a low value of absolute difference but it has a low possibility that it will occur in a real case scenario. Additionally, the accuracy of the prediction was very dependent on the variation of duty cycles [16][17][18]. Based on the observation, machines that were being operated at all times resulted in a low absolute difference value between the comparison of predicted and measured noise levels. Next, earth-moving machines were assumed to operate with full power at all times [7,20], but the operation cycle of earth-moving machines was inconsistent and tends to have a different duty cycle in reality [19]. Hence, the predicted value from simple prediction chart methods is rather higher than the actual noise levels. The inclusion of duty cycles in the simple prediction chart technique was suggested by previous researchers to improve the accuracy of the prediction [7].

Comparison of predicted results and actual measurement
Based on the data analysis, the range of average standard deviation for each of the case studies varied from 2.6 dBA to 8.1 dBA. The standard deviation in the simple prediction chart technique is to explain the noise level distribution within the sub-area. The average standard deviation of all the case studies is tabulated in Table 4. Case study 5 had the highest average standard deviation of 8.1 dBA due to the width to depth ratio of 1:8 (15 m: 120 m). When a sub-area has a greater depth, the range between the maximum and minimum noise levels tends to be larger because of inverse square law [7]. Nonetheless, the reason that caused case study 5 to have a high standard deviation value is due to the inconsistency of duty cycles and dynamic properties of three machines during the road construction activity. On the contrary, case study 1 with an aspect ratio of 1:1 had the lowest average standard deviation of 2.6 dBA which fulfilled the statement of sub-area with greater width will have a smaller standard deviation value because the noise level distribution is more well-proportioned [7]. Moreover, the duty cycle of case study 1 was fully operating throughout the measurement. For case studies 2, 3, 4 and 6, the average standard deviation values were 4.0 dBA, 3.6 dBA, 4.6 dBA and 3.0 dBA. Next, R-squared was applied in this study to evaluate the reliability of the predicted results. The results from Table 4 show the lowest R-squared value of 0.845, which is very reliable in this study. The highest R-squared value lies between case studies 1, 2, 3 and 6 whereas case study 4 has the lowest R-squared value of 0.845 followed by case study 5 with the R-squared value of 0.980. In summary, the simple prediction chart method is capable of providing good reliability in noise level prediction with an R 2 value ranging from 0.845 to 1.000 that indicated strong strength of association [28].

Conclusion
In conclusion, this study validated the predicted and actual noise levels from different construction activities. Actual noise levels measurement was conducted following BS 5228-1:2009 and BS ISO 6395:2008. The predicted noise levels were computed correctly by using the simple prediction chart technique. The data validation for this study discovered high accuracy and reliability results. To support the statement, the highest absolute difference value was 2.9 dBA accompanied with the highest relative error of 4.0 %. Furthermore, the highest mean absolute difference value was 2.0 dBA. The highest average standard deviation was 8.1 due to the aspect ratio. However, the lowest R 2 value was 0.845 among the case studies which indicates strong strength of association. Hence, the simple prediction chart method shall be adopted in the construction industry as a managerial and planning tool.

Declaration of conflicting interest
The author(s) declared no potential conflicts of interest regarding the research, authorship, and/or publication of this article.