This report summarizes noise modeling analyses conducted by David Coate Consulting (DCC) to determine the cause of increased highway noise levels experienced in the Fox Point community adjacent to relocated I-195 in Providence, Rhode Island. A number of possible causes exist, including increased vehicle volumes and speeds, transverse tines on new concrete bridge surfaces, and decreased shielding from topography/buildings caused by the highway relocation itself. In addition to diagnosing the cause of increased noise levels, this study also evaluates potential noise mitigation methods, the amount of potential noise reduction, and number of buildings benefited.
2. Community Noise Monitoring
David Coate Consulting (DCC) conducted a noise monitoring program at four residential locations in the community adjacent to Interstate-195 from May 6, 2011 through May 13, 2011. The purpose of the noise monitoring was to quantify noise levels from I-195 under various wind/weather conditions and for comparison with noise modeling results. Additional noise monitoring at Site 1 was undertaken from August 5, 2011 through August 10, 2011 in order to capture the noise effects of warmer weather and winds out of the south. Automatic noise monitors programmed to report one hour Leq (Level Equivalent--essentially hourly average noise levels) were employed and hand held short term noise measurements were also conducted to determine dominant noise sources as well to determine variations in noise level in the vicinity of each measurement site.
2.1 Day Night Average Noise Levels
Table 2.1 shows the results of the noise monitoring expressed in terms of Day Night Average Noise Level (DNL). The range of measured DNL values during the week of monitoring is shown by measurement site.
Figure 2.11 shows the locations of sites 1 through 4.
Figure 2.12 shows typical DNL values for residential areas.
Measured noise levels at Site 1 and 4 would be characterized as “Urban Residential,” Site 2 as “Very Noisy Urban Residential,” Site 3 as “Small Town Residential.” The wide variation in noise levels between Sites 1 – 4 is mainly due to different proximities to I 195. Numerous federal and state agencies as well as standards organizations such as the American National Standards Institute (ANSI) consider 65 DNL as the dividing line between acceptable and unacceptable noise environments for residential locations. Based on these particular noise measurements, Site 2 would fall into the “unacceptable” category.
2.2. Hourly Noise Level Time Histories
Figures 2.21 through 2.24 show measured hourly Leq values at each measurement site from May 6 through May 13, 2011.
2.3 Statistics/Sample Duration
The number of data samples collected is of interest in environmental noise monitoring because the goal is to determine the long-term average noise level at a particular site. The issue of adequacy of the number of data samples is of particular importance in this case because of the goal of capturing the noise effects due to a wide variety of weather/meteorological conditions. In general, the larger the variance in the noise level at a particular site, the longer the sample period has to be. Since a 3 dBA change in noise level is barely perceptible to most people, that threshold is typically used to define the 90% confidence intervals in environmental noise statistical analysis. That is, an adequate number of samples have been collected once it has been demonstrated that ± 3 dBA 90% confidence intervals have been achieved.
A statistical analysis was performed on the noise level data and the results are shown in Table 2.3. The 90% confidence intervals are all well below ± 3 dBA which indicates that a sufficient number of hourly samples were collected at each site. Further analysis indicates that a single twenty four hour period would have easily sufficed for statistical reliability at each of these measurement sites.
2.4 Weather Data
Table 2.4 shows wind direction, wind speed, and temperature for the Fox Point area during the monitoring period. As can be seen from this table, winds were out of the southern direction only on 5/6/11 and 5/7/11. On these two days, highway noise was more noticeable at certain noise monitoring locations.
Noise levels ranged 4 to 7 dBA higher on these two days at the four monitoring sites than during the rest of the week. The high end of the range (7 dBA) occurred at Site 1, Transit and Thayer Streets. A number of variables affect highway noise levels including traffic volumes, speeds, truck/car mix, and weather conditions. Local noise sources also affect noise measurements at each site.
In order to supplement the May 2011 noise measurements and more fully capture the effects of winds out of the south, additional noise measurements were collected at Site 1 from August 5 to August 10, 2011. Noise levels ranged from 55 to 60 DNL over this time period which was slightly noisier (by a maximum of 2 dBA) than during the May 2011 monitoring period.
3. Tire/Pavement Noise
3.1 Tire/Pavement Acoustical Testing (Wayside)
DCC performed noise measurements and analyses to quantify the noise characteristics of different pavement types and texturing along the relocated I-195 project.
On December 10, 2010 DCC and Maguire staff conducted noise measurements to determine the noise characteristics of the transverse tined PCCP pavement and Friction Course pavement. Please note that a few international measurement standards have been developed such as the On-board Sound Intensity (OBSI) method which incorporates two phase-matched microphones mounted near the tire/road surface on a test vehicle. Because of the costs and practicability associated with such methods, the current study employed a simpler approach which has some inherent limitations.
Noise measurements in 1/3 octave bands were conducted on the west-bound side at three positions:
Position 1: PCCP transverse tined at Station No. 271, see Figure 3.11
Position 2: Friction Course at Station No. 292.5, see Figure 3.12
Position 3: Friction Course at Station No. 309, see Figure 3.13
Noise measurements, vehicle volumes, vehicle classifications, and speeds were documented by lane for the half hour noise measurement periods. Measurements were performed at Position 1 for both half hour periods. Measurements were performed at Position 2 for the first half hour and Position 3 for the second half hour. Positions 2 and 3 were selected to determine how consistent noise levels would be at different locations with the same setback distance, compared to the results from Position 1. Figure 3.14 shows the results of these tests. Several initial observations include:
• PCCP transverse tined levels were consistently at 85 dBA
• Friction Course levels were also consistent at 72 to 74 dBA
• The difference between the pavements was 12 dBA, which is a very large difference. A 10 dBA increase is considered a doubling of perceived loudness.
• A tone at 1000 Hz is very prominent at Position 1 which is caused by the spacing and depth of the tines. Tonal noise can be annoying which may not be fully accounted for by A-weighting.
• The peak at 63 to 100 Hz is likely due to truck engine/exhaust noise which appears to be consistent between Positions 1- 3 (because the same traffic passed by these positions).
• The peak at 20 Hz in the Position 1 data is likely due to the bridge expansion joint near the microphone position. Truck impacts at that joint likely caused this peak, which does not affect the A-weighted values.
3.2 Traffic Noise Model Analysis
The results of Positions 1, 2, and 3 are not necessarily directly comparable even though the traffic was reasonably consistent. One of the main reasons for this would include the fact that roadway geometry is different between these sites.
For example, westbound ramps at Position 1 are elevated and therefore the line-of-sight to traffic is somewhat blocked for traffic on those ramps. In addition, ramps and mainline roadways are somewhat divergent at that position which differs from Position 2 and 3.
Another issue is that published PCCP transverse tine noise measurements show higher noise levels than that of the Friction Course pavement, but are in the range of 7 dBA higher. Consequently, the measured 12 dBA difference is suspect.
For these and other reasons it was important to also employ computer modeling using FHWA’s Traffic Noise Model (TNM) to determine basic site biases which could influence the pavement noise measurements.
Roadway geometries, by lane and elevation, along with traffic data were coded into TNM to accurately emulate the conditions during each half hour measurement segment.
Table 3.2 shows the results of the TNM runs compared with the noise measurements:
TNM results for Position 2 and 3 are in close agreement with measured results, but are grossly under predicted for Position 1. The TNM predicted value for Position 1 is consistent with the TNM results of Positions 2 and 3 (given the same traffic, speeds, and microphone setback distance), but is slightly lower presumably because of the aforementioned differences in roadway geometry. Because TNM assumes an “average pavement” type which closely matches the measurements at Positions 2 and 3, the under predicted TNM result at Position 1 indicates that an inherent site bias at Position 1 is not likely responsible for the 12 dBA difference between pavement types.
3.3 Other Pavement Noise Analyses
DCC also conducted informal “time history” noise tests inside a vehicle while driving over the two different pavements at a consistent speed. Even though these particular test results cannot be directly compared with exterior noise levels, they consistently show significantly higher “inside vehicle” noise levels on the PCCP traverse tined pavement compared with the Friction Course pavement.
The tone at 1000 Hz associated with the tining on the PCCP is directly related to vehicle speed and tine spacing. Wider spacing would result in a lower frequency and narrower spacing would result in a higher frequency. For example, a vehicle traveling at 60 mph with 1” spaced tines would result in a tone at 1056 Hz, consistent with the noise measurement data. This type of information can be used to calculate the effects of changes in tining and texturing.
3.4 Pavement Noise Study Literature Review
A substantial amount of work has been done internationally on the subject of tire/road noise and the effect of pavements/texturing. The following does not represent a full literature review, but includes the following highlights:
• Colorado DOT Tire Pavement Noise Study (2004). Used CPX test method. PCCP transverse tined 102.6 dBA, Friction Course 95.3 dBA (7.3 dBA difference). PCCP longitudinally tined 98.6 dBA.
• Caltrans Tire/Pavement Noise Study. Used OBSI test method. PCCP transverse tined 104 dBA, PCCP longitudinally tined 102 dBA, PCCP diamond grind 99 dBA.
• Evaluation of U.S. and European Concrete Pavement Noise Reduction Methods (2006). Random spacing of tines can eliminate/reduce noise issue. Diamond grinding texturing also reduces noise.
• Rubberized pavements are quieter than PCCP, but may not have longevity.
• Pervious pavements can absorb noise, but pores can be clogged over time, thus affecting the noise reduction performance.
• Numerous other pavement studies with PCCP transverse tining show the 1000 Hz peak, due to tine spacing of approximately 1 inch.
4. Washington Bridge Noise Measurements
“Near-field” noise measurements of the transverse tines of the concrete (PCCP) Providence River bridge surface had been previously conducted on 12/10/2010. Those measurements revealed that tire/pavement noise from the PCCP transverse tined pavement on the bridge was 12 dBA higher than that of the friction course (FC) pavement. In addition, a prominent peak at 1000 Hz was measured and corresponds with a very discernable (and objectionable) tone centered at that frequency. This tone corresponds precisely to the spacing of the tines and speed of vehicles.
The eastbound lane of the Washington Bridge also consists of PCCP transverse tined pavement, while the westbound lane consists of FC pavement. In order to characterize the differences between these pavements, noise measurements were conducted on 6/6/2011 simultaneously adjacent to the eastbound lane at the bridge (from the pedestrian bridge) and the FC section immediately to the west of the bridge.
Figure 4.11 shows the two noise measurement positions. Position 1 corresponds with the FC pavement and Position 2 corresponds with the PCCP transverse tined pavement. The noise measurements revealed that the PCCP pavement at this particular location is 5 dBA noisier than the FC pavement. Figure 4.12 shows the PCCP pavement results compared with those of the FC pavement. While there is a similar peak at 1000 Hz as measured from the Providence River Bridge, the peak is not as pronounced possibly due to a shallower depth of the tines.
Additional noise measurements of Washington Bridge were done on 8/5/2011 and 8/10/2011 which show smaller differences in noise levels between the two pavement types. It appears that slower traffic on those days may have caused tire/pavement noise to be lower and therefore engine noise was more predominant (i.e. which is unaffected by pavement type).
5. Noise Modeling
Noise modeling is important for this project for a number of reasons including identifying which buildings are affected by PCCP tine-generated noise as well as the magnitude of those noise levels compared with FC-generated noise levels. In addition, modeling is useful to determine the benefit of mitigation in terms of number of buildings which would experience a 5 dBA (for example) reduction. This information in turn can be used to evaluate the “reasonableness and feasibility” of mitigation in terms of cost/building benefited.
5.1 CADNA/TNM Modeling
CADNA is an internationally recognized environmental noise computer program which is suitable for this modeling, particularly in terms of accurately determining the shielding effects of topography and buildings which is important in this case. However, the Traffic Noise Model (TNM) is Federal Highway Administration’s (FHWA) sanctioned noise model for federally funded highway projects in the U.S. Consequently, DCC used TNM to determine basic “source” noise levels for each modeling scenario which were then imported into CADNA. TNM inputs include roadway geometry in terms of plan and profile, vehicle speeds, and volumes of automobiles, medium trucks, and heavy trucks.
Table 5.1 shows the traffic data used in TNM for each noise modeling scenario.
5.2 2011 Am Peak
The results of TNM for the 2011 Am Peak scenario were imported into CADNA along with building footprints and building elevations. In addition, noise levels and frequency spectra associated with the PCCP transverse tined pavement and FC pavement were input into the model. The resulting CADNA noise contour map associated with this scenario is shown in Figure 5.21
The large contour lobes extending from the bridge sections show the effects of the noisier pavement type on certain portions of the Fox Point community.
Table 5.2 shows a comparison between CADNA predicted and measured noise levels at each of the four community noise monitoring sites. The measured noise levels shown here are arithmetic average hourly Leq values over the entire measurement period (as opposed to DNL values). There is no reason to expect that predicted and measured values would be in close agreement because the traffic data used for the predictions was peak hour which would probably be high compared with twenty four hour average traffic data. There are many other factors which would explain differences between modeled and measured including the effects of local ambient noise sources which were not included in the modeling. However, as can be seen from this data, predicted values are reasonably close to measured values.
5.3 Old Highway Am Peak
TNM source noise levels associated with the Old Highway Am Peak scenario were imported into CADNA and the results are shown in Figure 5.31. Noise levels are lower than those associated with the 2011 Am peak scenario for a number of reasons including lower vehicle speeds and shielding due to topography. For example, the old highway was depressed relative to the community in the western part of Fox Point. The new highway is further away from the community in that section, but is elevated so that many buildings have a more direct line-of sight to the PCCP transverse tined pavement.
5.4 Changes in Noise Level Due to the Relocated Highway
Figure 5.41 shows the difference between the data in Figure 5.21 and Figure 5.31 which reveals the areas which have experienced 10 dBA or more increases due to the relocated highway –based on the modeling results. This analysis shows that certain areas are more affected than others due to the aforementioned changes in noise levels. The increases in noise level shown much further from the highway are artifacts of the modeling due to the fact that local ambient noise sources are not included in the model. That is, as highway noise levels decrease with increasing distance from the highway, local ambient noise levels would begin to dominate which in turn would cause those noise level increase contours to disappear.
The reasons for increased noise levels due to the new highway are complex and involve changes in both traffic conditions and changes due to decreased acoustical shielding. With respect to changes in traffic conditions, traffic is highly variable and the long-term noise effects only due to changed traffic conditions are not immediately obvious. To address this, DCC modeled the old highway assuming that traffic conditions were identical. In other words, the old highway was modeled using the 2011 Am Peak traffic data. This modeling exercise is obviously not realistic because the old highway would not have been able to accommodate these higher traffic volumes and speeds. The resulting noise contour map looks similar to Figure 5.31 except that noise levels are somewhat higher due to increased vehicle speed. Figure 5.42 shows the differences between this scenario and Figure 5.21. Note that the differences are shown as smaller increases of 3 dBA or more.
The analysis shown in Figure 5.42 shows that there are several areas in the western part of Fox Point which likely experienced a 3 to 5 dBA increase in noise level just due to the change in highway alignment, assuming traffic conditions were identical. This is a small increase in noise level, but would be noticeable to most people. However, as previously discussed, increases in noise levels in this situation are likely due to a combination of the PCCP tined pavement, improved line-of-sight from the highway to the residents, and increases in traffic volumes and speeds.
6.1 Mitigation Options
When noise mitigation is warranted, FHWA typically employs noise barriers which reduce the line-of-sight from noise source to listener and can provide 5 to 10 dBA of noise reduction. In a few isolated cases building sound insulation, which mainly involves the use of acoustical replacement windows and doors, has been used. Building sound insulation only improves interior noise levels, and is typically not cost effective for large numbers of buildings. In this case, noise barriers are not feasible because they would have to be placed on the edge of the bridge structures to be effective. This approach is not structurally feasible. A far more effective solution would be to eliminate the transverse tines which are the main cause of the noise issue.
6.2 Pavement Re-texturing/Overlays
Several state DOT’s (for example Arizona, California, and Colorado) have dealt with the PCCP transverse tined pavement noise issue by re-texturing the pavement or overlaying the pavement with another pavement type such as asphalt rubber friction course pavement. In general, asphalt pavements are initially quiet (more so than un-tined PCCP pavement), but gradually over time become noisier as the once sound-absorbing voids in the pavement become clogged with debris.
On-board Sound Intensity (OBSI) measurements employs two phase-matched microphones mounted on a car near the pavement/tire interface to accurately measure pavement/tire noise and importantly to allow direct comparison of different pavement and texture types. OBSI measurements have not yet been conducted for the I-195 project, but are being planned in order to more accurately quantify the noise emission of the PCCP transverse tined pavement for comparison with other pavements/textures. This is a critical step for this project, in order to accurately determine the amount of noise reduction achievable. In turn, that data would be re-run in the CADNA model to determine the net effect in the community which depends not just on the PCCP pavement sections but also the other non-PCCP sections as well.
Numerous studies using OBSI techniques have measured the results of pavement grinding with a wide range of reported noise reduction values. Reported noise reductions of diamond ground transverse tined pavements range from 3 dBA up to 12 dBA . OBSI measurements of the existing transverse tined pavements on the Providence river and Washington bridges will allow direct comparison to other published data and more precise estimates of actual noise reduction values.
It is important to note that the percentage of heavy truck traffic can influence the effectiveness of a quieter pavement surface. Depending on vehicle speeds, the source of noise of heavy trucks is the engine and exhaust which is not affected by pavement type. Consequently, at higher heavy truck volumes, the quietness of the pavement may not matter as much.
Figure 6.3 shows modeled noise levels of the 2011 Am peak scenario assuming that noise reduction values near the top end of the range (11 dBA) would be achieved by diamond grinding the transverse tined pavements.
Figure 6.3 can be compared directly against Figure 5.21 which shows substantial noise reduction especially in areas of Fox Point in the vicinity of the PCCP transverse tined pavement. Figure 6.4 shows the noise level differences between these two datasets.
This figure shows improvements due to resurfacing at 5 dBA or more, which would be a very noticeable improvement. A 5 dBA noise reduction is considered substantial, but the removal of the tone would be very noticeable, which is not completely accounted for in the A-weighted noise reduction value. A-weighting generally does not account properly for the additional annoyance due to tones. The western section of the Fox Point area would be benefited the most by resurfacing.
7. Feasibility And Reasonableness of Mitigation
7.1 FHWA Policy
FHWA noise projects typically evaluate noise mitigation on the basis of feasibility and reasonableness. “Feasibility” refers to the engineering practicality of the mitigation and in general sets a minimum standard of a 5 dBA noise reduction. This minimum noise reduction value is employed because in general lower noise reductions may not be noticeable. “Reasonableness” refers to the cost of the mitigation relative to the amount of noise reduction and number of residences benefited. In this case, it is possible to evaluate feasibility and reasonableness of pavement resurfacing on the basis of the data shown in this report. However, such cost/benefit analyses are normally applied to noise barriers and may not be completely applicable with respect to pavement modifications.
7.2 RIDOT Noise Mitigation Policy
RIDOT defines “reasonableness” in terms of the cost per protected dwelling unit (DU) to be less than $25,000. A protected dwelling unit is defined as one receiving at least 5 dBA noise reduction. An alternative definition is that the cost/dBA/DU should be less than $2500.
The improvements shown in Figure 6.4 are somewhat misleading because the analysis does not take into account the fact that local ambient noise sources limit the distance away from the highway at which highway noise is audible and therefore where improvements due to mitigation would be realized. The approximate setback distance and location from I195 where improvements would be realized is along Transit Street and in some areas a few blocks further from the highway. Using that information and GIS data, Figure 7.2 shows the buildings (highlighted in red) which would experience a 5 dBA or greater improvement due to resurfacing/retexturing.
According to the GIS data, an estimated 567 buildings would experience a 5 to 11 dBA improvement due to resurfacing/retexturing. The approximate cost for diamond grinding has been estimated to range from $600,000 to $900,000. This results in a cost/DU of $1587 which is less than RIDOT’s ceiling of $25,000/DU. Assuming the high end of the range of improvement due to diamond grinding, this analysis shows that this noise mitigation method would be feasible and reasonable according to RIDOT criteria.
Assuming the low end of the range of improvement due to diamond grinding (3 dBA), CADNA was re-run with those assumptions and the results were imported into GIS software. The GIS analysis shows that 517 buildings would experience at 2-3 dBA improvement. This analysis indicates that the diamond grinding would fall below RIDOT’s feasibility criterion given these assumptions (i.e., less than 5 dBA improvements per DU). On the other hand, the removal of the tone due to the tines would likely be very noticeable which may not be fully accounted for by the RIDOT feasibility criteria. Assuming that a dwelling unit could be considered “protected” with only a 3 dBA improvement (i.e., due to the removal of the tone), the cost/DU/dBA would be $580 which would satisfy RIDOT’s reasonability criterion.