Field Study of the Noise Exposure Inside Running Metro Unit

The noise levels inside metro units are considered a significant problem that makes passengers suffer from severe damage, especially for those who use the metro periodically. This research evaluates the acoustic environment inside the metro car and studies factors affecting the noise levels inside metro units and developing models for estimate noise in the metro unit while moving between stations. Greater Cairo Metro (GCM) Line 1 has been selected as a case study. A sound level meter was used to measure the equivalent sound level in dBA and evaluate the noise inside metro units. The results indicate that the noise levels are unacceptable compared with the international noise exposure standards. The highest measured noise level inside metro units is 91.2 dBA. These unacceptable noise levels led to more investigation of factors that may affect noise levels inside metro units. Other data have been collected, such as the speed of the train and the track alignment details. The results showed that the noise increases with the increase of the train speed until the speed reaches a specific value, then it decreases depending on the maintenance status and the train type. In addition, the noise levels through curved underground tracks are higher than the levels along straight surface tracks by 18 dB(A).


Introduction
Metro is an essential component of the public transit system in urban facilities [1]. Following the practical urban progress of all countries, rail transit becomes a major urban transit mode in developed countries for its advantages such as mass transit capacity, safety and reliability, fast speed, and comfort. Metro is believed to be more environmentally friendly than other public transport modes. Also, it reduces the number of cars on the road, which in turn reduces the air pollution and traffic congestion caused by individual vehicles. For these advantages, big cities have a considerable scale of rail transit routes which have gradually become the first choice of transport tool by the majority of the citizens [2,3]. Metro becomes the key point of urban transportation development strategy due to its social and economic properties and is regarded as the backbone of the urban transit system. Despite all the mentioned advantages, noise exposure due to transportation noise as railway and traffic noise has been reported to affect millions of people worldwide and has been related to various health-related issues [4,5].
There is a lot of research and information about the noise impact on hearing health [6]. The World Health Organization (WHO) acknowledges environmental noise as one of the top environmental risks for health and wellbeing. Noise exposure can lead to various effects depending on noise level and duration [7]. Noise-induced hearing loss is irreversible damage to the ears caused by exposure to high noise levels [8]. Exposure to high levels of noise has adverse effects in the short and long term. The long-term effects include damage to human health, as shown in Figure 1. It is feasible that repeated exposure to high noise levels at railway-centered stations may damage the hearing and circulatory systems of transit riders [9][10][11]. On the other hand, for the short-term effect, the high levels of noise on station platforms create an unpleasant environment for passengers waiting for their train. Riders have difficulty holding conversations with fellow passengers or on their phones. Research into annoyance caused by noise shows that people exposed to high noise levels have problems in concentration, making even silent activities, such as reading problems [12,13]. However, little is known regarding this effect.

Figure 1. WHO pyramid of health effects of noise
Noise is considered an "unpleasant sound" [14]. Noise can be described in terms of three variables: amplitude (loud or soft), frequency (pitch), and time pattern (variability). These combine to create the intensity of the noise [15]. The most crucial noise source from railways is the rolling noise caused by wheel and rail vibrations induced at the wheel/rail contact, resulting from increased speed, braking system, and neglect of the periodic maintenance for rolling stock or rail elements [16]. There are other sources of railway noise: High alarms are used on platforms and while meeting another train in the opposite direction.
The railway car's noise can irritate passengers, especially when windows are open while the train moves inside the tunnel [17]. Noise can reach commuters inside the metro units in three ways [18]. Figure 2 shows the three ways noise can get commuters inside metro units: 1) External airborne sound and flow turbulence cause pressure fluctuations on the car shell, which transmits a portion to the interior. 2) structure-borne vibration is transmitted from wheels, motors, and under-car equipment along solid paths to interior surfaces, which vibrate and radiate noise inside.3) Airborne noise from the interior [19]. The primary parameter for the Sound Pressure level (SPL) is an A-weighted equivalent continuous sound pressure level or LAeq [20].

Figure 2. The three ways noise can reach commuters inside metro units
A study by M. A. Abdalla conducted a field survey to compare the statistical noise levels released from buses and some of the GCM Line1 stations and the Tram. Noise levels measured onboard exceeded the maximum allowed noise of 68 dB(A), where the equivalent continuous noise level LAeq measured lay within the range of 70-77.5 dB(A). The Max level LAmax reached 90 dB(A) due to the use of alarm signals with a duration of 3 seconds. When the train is in the tunnel, the measured noise level onboard was more significant than the noise level at the surface (at-grade). This is caused by reflections of sound at the subway wall, which increase noise levels [21].
Another study conducted by Mostafa E Aly investigated GCM noise problems and related health hazards when the noise is measured on the station platform and inside metro units. A comparison has been conducted between different noise parameters and international criteria to show that noise levels in underground stations of the GCM line 2 are rejected [22]. The international standards used in comparisons were adopted by the US Department of Housing and Urban Development [23,24]. Measurements of equivalent sound level LAeq in dB(A) are taken at each platform. LAmax ranges from 87.8 to 102.5 dBA. The max value of LAmax was shown between El-Gamaah and El-Behoos stations, and this is due to the tunnel entry. L99 is in the range of unacceptable according to the criteria adopted by USHUD and the noise pollution index (LNP), calculated from the measurements of noise levels measured in dBA.
A study by Shimokura and Soeta has indicated that the LAeq of train noise in underground stations is 5 dB higher than that of in-ground stations. Because the walls and the ceiling of the underground stations are regularly covered with reflective fire-resistant materials (e.g., steel or vitreous enamel panels), sound reflections and reverberations lead to an increase in train noise. The LAeq of train noise has been different by platform types. The LAeq of train noise on island platforms was 2.3 dB higher than that inside platform [25].
Another study by Shimokura and Soeta has indicated that the A-weighted sound pressure levels in the underground stations are 6.4 dB higher than that in the surface stations. For the surface stations, the averaged LAeq and IACC on the side platform are 3.3 dB larger and lower than those on island platforms. For the underground stations, the averaged LAeq on island platforms is 3.3 dB higher than that on side platforms when a train comes into the station. The LAeq at the entrance and exit ends of the platform are high when a train comes into and goes from the station [26].
Even though passengers take more time inside the metro units, it hardly could be found in the literature. The contributions of this study are to assess the noise levels inside metro units and develop new relations between noise levels and the speed of trains regardless the train's type. Moreover, the factors that may affect the noise levels are investigated.

Research Methodology
The methodology is divided into three parts. The first one discusses the workflow of the research. Then, the second one illustrates the needed data and the process of data collection. Finally, the third part describes the case study region.

Research Workflow
The workflow is depicted in Figure 3, where the measurements of noise levels have been made inside the units while traveling the units to evaluate the noise problem and to suggest solutions to the problem followed by the data analysis. Then, a comparison between Sound Pressure Level (SPL) inside the metro units between stations is conducted with the international standards of noise exposure. Also, factors (train speed, horizontal and vertical geometry, the type of train, tunnel, or surface stations) that may have noise effects have been tested.

Data Collection
Three datasets were collected for this study, including sound level measurements, train speed data, and alignment. The first dataset was the measurements of the equivalent sound level in dBA, Leq, which have been made inside the metro units between line 1 stations starting from Marg to Ain Helwan station for the three-train type during off-peak hours to avoid commuter noise. These data were crucial in noise assessment in Line 1 and to conduct a comparison with the international standards of noise exposure. For this dataset, the measurements of the equivalent sound level in dBA, Leq, which has been made inside the metro units between line1 stations sound level meter tool has been used. A sound level meter is used for sound measurements. It is generally a handheld instrument with a microphone. The best type of microphone for sound level meters is the condenser microphone, which combines precision with stability and reliability. The diaphragm of the microphone responds to changes in air pressure caused by sound waves. That is why the instrument is sometimes referred to as an SPL Meter. In this study, the data has been collected using Sound Level Meter UT352.
The speed of trains between the stations was measured using the Androsensor mobile application. This type of data was used to test the effect of the train speed on the noise level. Finally, the geometrical alignment of Line 1 was used to study the impact of the curved track and type of station (underground or surface) on the noise level. Figure (

Noise-speed Estimation Model Development
Linear regression is a common simple modeling technique that has been used in all fields of science and engineering. Here, this technique is used, and the models were validated by the R-squared and the Mean Absolute Percentage Error (MAPE) as Goodness of Fit measures (GOF).

Data Description
Due to the variation in the type of rolling stock in GCM, three types were classified for data measuring. Table 1 shows the specification of these three types. For each type, SPL has been recorded using a sound level meter along line1 inside the metro units as mentioned previously. Table 2 shows a sample of the Min, Max, and Mean measured SPL along line 1 inside metro units for type C.

Determination and Assessment of Noise Inside the Metro Units
For the noise determination and evaluation, LAeq, LAmax, LNP, and L99 have been calculated and then compared with the international standards of noise exposure shown in Table 3. LAeq, LAmax, LNP, and L99 can be defined as follows: LAeq: A-weighted equivalent continuous sound pressure level during a single noise event, Leq can be described mathematically by the following equation [27]: Where Leq is the equivalent continuous linear weighted sound pressure level re 20µPa, determined over a measured time interval Tm (sec); P(t) is the instantaneous sound pressure of the sound signal; P0 is the reference sound pressure of 20µPa; LAmax: maximum sound pressure level during a single noise event; LNP: the noise pollution index that considers the deviation of the readings from the mean. It can be calculated from Equation 2 [28].
Where; σ = the standard deviation of the readings; L99: is the background noise level, which represents mostly the lowest noise level during a single noise event (i.e., L99 represents the noise level exceeded for 99% of the measurement period).

Effect of Speed
The second objective of this research is to study the effect of train speeds as a crucial parameter affecting the noise level. For the three types, the measured SPL and train speed between two surface stations with no curved track was recorded. The purpose of that is to avoid the noise produced due to horizontal alignment and to the amplification of noise through tunnels. So, the selected data for this issue were taken between Hadayaq El-Maadi and Maadi stations. The distance between these two stations is 1.25 km. Figure 5 shows the chosen track between Hadayaq El-Maadi and Maadi stations for a speed noise study. Also, the speed profile has been divided into two parts; the acceleration part and the deceleration part. The purpose of this division is to avoid the noise produced from braking as its effect takes place during deceleration.

The Effect of Horizontal Alignment and Station Types on Noise
Finally, the third objective of this research is studying the effect of horizontal alignment and station types (surface or underground) on noise levels. To study these two parameters, the following measurements were taken into consideration to select the data suitable for this issue:  Noise levels between two surface stations without any horizontal curves;  Noise levels between two surface stations with horizontal curves;  Noise levels between two underground stations without any horizontal curves;  Noise levels between two underground stations with horizontal curves; Table 4 shows more details for the selected tracks.  Figure 6 shows the maximum noise level for the three types along line 1 compared with the international standards of maximum noise exposure. The following can be noticed from this figure:

Assessment of Noise Inside the Metro Units
 For type A, the lowest and highest LAmax were 78. 24 Table 3;  Type B was the worst type in terms of L99 as it exceeds 68.00 dBA, which is unacceptable;  Type A was better than type B, but the recorded L99 exceed 68.00 dBA, which is unacceptable. Also, it can be noticed there is no large fluctuation in the recorded L99;  Type C was the best in terms of L99.  Clearly acceptable ( L < 35 )

Normally acceptable ( 35 < L < 53 )
 Type A was better than type B, but the recorded LNP exceeded 73.00 dBA, which is normally unacceptable. Also, it can be noticed that there is no large fluctuation in the recorded L99;  Type C was the best in terms of LNP.    Table 3, LAmax is acceptable within the range of 63.5dBA and 73.5 dBA for train movement along a straight surface track. Type C did not exceed this limit, while type B has exceeded this limit at a speed of 25 km/hr. Besides, type A has exceeded the allowed limits at the beginning of the train motion.

A Speed Impact Assessment
The linear regression analysis has been used to calibrate and validate models that describe the speed-noise relationship through the train acceleration and deceleration. The following table shows the calibrated models and the coefficient of determination for each model. Where SPL = sound pressure level in dB(A) and S = speed of the train in km/hr. Table 5, the R-squared values for model C are all above 0.80, which reflects that model C fits the data to some extent. Also, the R-squared values for models A, B are less than 0.60, which reflects a relatively poor prediction power of the models. MAPE has been used for validating type c models. MAPE values were 2.08 and 2.97 for acceleration and deceleration models, respectively. Figure 12 shows the noise levels profile between two surface stations with a straight track (Hadayeq El-Maadi /El-Maadi) compared with the noise levels profile between two underground stations with a straight track (Orabi /Nasser), while Figure 13 shows the noise levels profile between two surface stations with a straight track (Hadayeq El-Maadi /El-Maadi) compared with the noise levels profile between two surface stations with a curved track (El-Demerdash /Ghamra). The following can be concluded from the two figures:  The noise levels between underground stations with the straight track are larger than those between surface stations with a straight track.

The Effect of Horizontal Alignment and Station Types on Noise
 The noise levels between surface stations with the curved track are larger than those between surface stations with a straight track. to compare the noise levels between the mentioned stations in Table 4. Table 6 shows the values of these parameters. The following can be concluded from this table:  The lowest values were between surface stations with a straight track, while the highest values were between underground stations with the curved track.
 The values of these parameters between underground stations with the straight track were less than that between underground stations with the curved track, but at the same time, the values were higher than those between surface stations with the curved track.

Discussion
It is clear from the previous analysis and results that LAmax, L99, and LNP exceed the international standards of noise exposure with a range reach to (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18) dBA, especially in type A, B inside the metro units along line 1. The increase in noise levels, which is observed in types A and B, is due to the poor ventilation system of the units. This causes the passengers to open the windows to compensate for the required quantity of air. On the contrary, the noise levels of type C were much less than those of types A and B. This is because type C is the newest deployed type with a newer braking system and produces less noise. Also, the units are air-conditioned with a good ventilation system, and there no need for open windows. Also, the windows themselves are sound-proof windows, which prevent outside noise from entering the units. The closing of the side windows attenuates the noise inside the units by about 15 to 20 dBA.
In general, the noise levels of the three studied types are different. The noise levels are unacceptable compared with the international standards of noise exposure. The high noise levels make passengers suffer from severe damage, which may deteriorate especially for those passengers who use the metro periodically. The noise produced from one or more of the following sources: Aerodynamic and mechanical noises of the train, The train siren, The brakes when applied, Two trains passing by each other at the same moment, which is a frequent event, The passengers themselves and the recorded attention sound systems inside the units, The condition of the railway infrastructure is an important factor in the high noise levels in the three types, and The short distance between stations, which does not allow an increase in speed to reduce the equivalent noise during the trip between the stations.
The unacceptable noise levels compared with the international standards of noise exposure led to more investigation on factors that may affect noise levels inside metro units, such as the speed of the train, the horizontal alignment, and the type of station. Concerning the noise-speed relationship, the noise is directly proportional to the speed of the train regardless of the type of the train until the train reaches a certain speed, and then noise starts to decrease again. The cause of this phenomenon is that at high speeds, the friction decreases, and so the produced noise decreases as well. The speed at which the noise curve decrease is different from one type to another depending on the maintenance status and the train type.
Based on the results reported for the linear regression models in Table 5, model C is the best in terms of GOF when compared to other models. It can be concluded that noise is directly proportional to the train speed as shown in type C models, but the noise has been affected by some other factors rather than the train speed for type A and B models. As for type A and B models, the models are weak in describing the speed-noise relationship as more factors related to the train type itself affect noise levels that cannot be faded The second point is the speed at which the produced noise exceeds the international standards of maximum noise exposure. Those speeds are as follow:  For type C, the produced noise did not exceed the permissible limits along the study track;  For type B, the produced noise exceeded the permissible limits along the study track at a speed of 25 km/hr;  For type A, the produced noise exceeded the permissible limits along the study track since the train starts moving.
Type C has better performance than types A and B in terms of the noise produced due to the previously mentioned reasons. And so, it has been selected to study the effect of horizontal alignment and station types (surface or underground) on noise levels.
Regarding the effect of horizontal alignment and station types (surface or underground) on noise levels, it can be concluded from the analysis of this part is that the noise through underground tracks has the highest values. On the other hand, the noise along surface tracks has lower values. The reasons for these results are as follow:  The sound reflections and reverberations at the tunnel cause an increase in train noise;  During the train movement over the curved track, there is an additional force affecting the track. This force causes a much disturbance on noise levels higher than the noise levels produced by the movement on a straight track.

Conclusions
This paper assesses the noise inside the metro units for GCM as one of the most critical environmental issues. It discusses the effect of speed, rolling stocks type, and line alignment on noise level as the commuters spend most of the time inside the metro unit. The main findings from this research are summarized in the following points:  The noise levels are unacceptable compared with the international standards of noise exposure inside the metro units along Line 1.
 The highest noise levels recorded were for types A and B, due to poor braking system and poor ventilation system of the units, which causes the passengers to open the windows to compensate for the required quantity of air. This case allows the outer noise to enter the train units.
 The noise levels of type C were much less than those of types A and B for the following reasons. The first reason is that Type C is the newest deployed type with a newer braking system and produces less noise. The second reason is that the units are air-conditioned with a good ventilation system and the windows themselves are sound-proof windows. The closure of side windows attenuates the noise inside by 10 to 20 dBA.
 During acceleration, the noise is directly proportional to the train speed regardless of train's type. The noise increases with the increase of the train speed until the train reaches a certain speed. This speed is found to be 82.5, 78, 64.5 km/hr for types A, B, and C, respectively. After this speed limit, the noise starts to decrease.
 During deceleration and braking interval, the noise levels are high due to additional noise sources, such as the braking system and the railway infrastructure condition.
 For the train movement on a straight track, type C did not exceed the permissible limit. On the contrary, type B exceeded the permissible limits at a speed of 25 km/hr. For type A, the noise levels exceeded the permissible limits from the beginning of the train motion. For types A and B, the noise excess is within a range between (10-15) dBA.
 In general, the track alignment influences the noise level. The noise levels through underground tracks are higher than those through surface tracks by about (5-10) dBA.
 The highest noise levels are when the train moves through curved underground tracks followed by straight underground tracks, while the lowest noise levels are through straight surface tracks.
Based on the above results, it recommended to:  Chang the maintenance plans for both rolling stocks and infrastructure to be compatible with the international standards of noise levels.
 Construction specification must contain items for covering the tunnel walls with sound-absorbing materials to be used in noise abatement strategy.
 Upgrading the currently used system for alerting and guiding inside the train units and within the stations,  Using a ballast layer during the design of tracks to absorb and reduce the sound radiation from rails and sleepers, especially for underground tracks.