Supplemental Guidance on the Application of FHWA’s Traffic Noise Model (TNM) (2014)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

11 C H A P T E R 3 3.1 Introduction Four of the 16 research topics covered in this report are related in that they involve “interrupted flow,” that is, the decel- eration, acceleration, and possible stopping and idling of vehicles. The topics are signalized interchanges, intersections, roundabouts, and area sources. The first three topics are cov- ered in this chapter because they have many features and components in common. Guidance on modeling accelerating and decelerating traffic associated with the special situations of toll facilities, service areas, weigh stations, and park-and- ride lots is given in Section 4.2 of Chapter 4 of this report. 3.1.1 Signalized Interchanges The two main types of signalized interchanges addressed in this research are the traditional diamond interchange and the single point urban interchange (SPUI), shown in Figure 4. The main issue with interchanges of all types, including unsignalized ones such as clover leafs, is over-modeling or micro-modeling without any improvement in the accuracy of the results. The converse issue would be under-modeling and thus ignoring potentially important effects, particularly the effects of accelerating vehicles, where noise-sensitive receptors are nearby. Given that so much deceleration and acceleration occurs near the center of an SPUI, an important question is how much influence the interrupted flow has on overall levels at receivers near the interchange. This influence is more likely to be important at a diamond, where sensitive receptors could be very close to the changing-speed traffic. However, the influ- ence can be small depending on parameters such as the domi- nance of mainline traffic noise if the crossing road passes over the highway (so that the ramp embankments shield the main- line noise) and the proximity of the receptors to the ramps and the mainline. 3.1.2 Intersections This task area includes both signalized and unsignalized intersections. The main issue with intersections of all types is over-modeling or micro-modeling without any improve- ment in the accuracy of the results. As with signalized inter- changes, the converse issue would be under-modeling and missing potentially important effects on received sound levels when noise-sensitive receptors are nearby. 3.1.3 Roundabouts Roundabouts have become a more and more common design feature for regular intersections and for highway entrance and exit termini in the last 15 years. As noted in the Foreword of NCHRP Report 672: Roundabouts: An Informational Guide, Sec- ond Edition,5 there were only about 38 “modern” roundabouts in the United States in 1997, but by 2010, there were over 2,000. A modern roundabout differs from the older, large, high-speed rotaries still in use in the country and also from the smaller traffic circles typically used to calm traffic in suburban neigh- borhoods. As illustrated in Figure 5 (Exhibits 1-1 and 6-2 from NCHRP Report 672), a modern roundabout is characterized by • A generally circular shape with counterclockwise flow. • A single lane or multiple lanes with signing and pavement markings that eliminate the need to change lanes to exit from the roundabout. • Use of splitter islands between the approach and departure lanes that creates entry geometries that force slow speeds, positively direct the motorist in the correct direction, and provide a refuge for crossing pedestrians. Signalized Interchanges, Intersections, and Roundabouts 5 Rodegerdts, L. A., J. Bansen, C. Tiesler, J. Knudsen, E. Myers, M. Johnson, M. Moule, B. Persaud, C. Lyon, S. Hallmark, H. Isebrands, R. B Crown, B. Guichet, and A. O’Brien, NCHRP Report 672: Roundabouts: An Informational Guide, Second Edition, Transportation Research Board of the National Academies, Washington, D.C., 2010.

12 • Use of yield signs at the entries rather than stop signs or signals. The goal of modern roundabout design is to slow, but not stop, the vehicles, achieving smooth and safe functioning of the intersection. As with signalized interchanges and intersections, the main issue with roundabouts is finding a level of detail in modeling that achieves accurate results while avoiding over-modeling that does not improve the results. 3.2 Modeling Acceleration and Deceleration 3.2.1 Acceleration FHWA TNM computes the acoustical effect of accelera- tion as vehicles pull away from traffic-control devices such as stop signs, toll booths/barriers, and traffic signals, and also along highway entrance ramps. FHWA TNM calls roadways with traffic-control devices “flow control” or “interrupted flow” roadways. As vehicles accelerate on these roadways, vehicle reference energy mean emission levels (REMELs) are higher than the REMELs of cruising vehicles at the same speed, according to field research done as part of the FHWA TNM development.7 That research developed a “full throttle” emission-level database and speed-distance-grade algorithms Figure 4. SPUI.6 Figure 5. Modern roundabout characteristics.8 8 The schematics shown in Figure 5 are, on the top, Exhibit 1-1 (p. 1–3), and on the bottom, Exhibit 6-2 (p. 6–9), from Rodegerdts, L. A., et al., NCHRP Report 672: Roundabouts: An Informational Guide, Second Edition, Transportation Research Board of the National Academies, Washington, D.C., 2010. ©National Academy of Sciences 2010. 6 Imagery © 2014 Google, Map data © 2014 Google. 7 Bowlby, W., R. L. Wayson, S. Chiguluri, M. Martin, and L. A. Herman, Interrupted Flow Reference Energy Mean Emission Levels (REMELs) for the FHWA Traffic Noise Model, U.S. Department of Transportation, Research and Special Programs Administration, John A. Volpe National Transportation Systems Center, Report No. DOT-VNTSC-FHWA-97-1 and FHWA-PD-97-019, January, 1997. for computing the speed of vehicles as they accelerate along an interrupted flow roadway. Collection of deceleration data was not within the scope of that work. An interrupted flow roadway is designated by choice of a “Control Device” in the FHWA TNM Roadway Input dialog box. When this choice is made, the modeler-supplied speeds

13 on the traffic tab of the roadway input dialog box are treated by FHWA TNM as the final speeds that vehicles will try to reach during their acceleration. The modeler provides a start- ing speed for all traffic on the roadway, called the “speed con- straint,” which would be zero for a stop sign, but which could be non-zero for an entrance ramp, as an example. The mod- eler also provides a percentage from 0 to 100% for “vehicles affected,” which, in the case of a control device such as a traf- fic signal, results in a percentage of the traffic to experience acceleration, while the remaining percentage is modeled as cruising along the roadway at the modeler-provided speeds. The FHWA TNM Technical Manual 9 states that FHWA TNM computes accelerating speeds for each vehicle type along a road- way’s length as a function of roadway grade until the final speed is attained or the end of the roadway is reached. Research con- ducted under NCHRP Project 25-34 identified a problem with the condition of reaching the end of the roadway and proposes a solution. The FHWA TNM Technical Manual also notes that FHWA TNM tracks speeds from one roadway segment to the next for a given roadway, but not from one roadway to the next. 3.2.2 Deceleration FHWA TNM has no built-in function for modeling decel- eration. For deceleration conditions, NCHRP Report 311: Predicting Stop-and-Go Traffic Noise Levels10 gives a founda- tion from which to work. That research project developed a methodology for using the constant-speed STAMINA 2.0 program (FHWA TNM’s predecessor) in acceleration and deceleration situations that occur on interrupted flow facili- ties or on ramps. The work included measurement and analy- sis of both accelerating and decelerating vehicle sound levels and a series of sensitivity tests using STAMINA 2.0. NCHRP Report 311 defined two “zones of influence” (ZOIs) to represent the last two segments of a roadway being used to model deceleration, as illustrated in Figure 6. As shown in Table 3, which is modified from Table 7 of NCHRP Report 311, guidelines are given on the lengths of these segments, as a function of approach speed, and “equivalent speeds” to use for each vehicle type on each segment. Both the segment lengths and speeds were empirically derived from field measurements reported in NCHRP Report 311. An issue is that those speeds are based on the circa-1975 emis- sion levels in the STAMINA 2.0 model, not the circa-1994 emission levels in FHWA TNM. 3.2.3 Questions on Modeling Acceleration and Deceleration For acceleration, as noted above, FHWA TNM includes REMEL equations and changing-speed algorithms that were derived empirically from field-measured sound-level and speed data. There was no identified or expected need to question that data or collect new data. There are several issues with modeling an acceleration: • What roadway segment length should be used to minimize errors in the predicted level? • What starting speed, or Speed Constraint, should be used for an acceleration roadway? • Where should the interrupted flow roadway start in the pres- ence of a queue of traffic at a signal, whether that signal is at the end of a ramp or at the intersection of two through roads? • How does the choice of percentage of traffic to use for the Vehicles Affected parameter affect the results? • To what level of detail should the traffic movements be modeled? • How far should an acceleration roadway such as an entrance ramp be extended to properly account for the noise from the accelerating traffic, especially heavy trucks? The issues for modeling a deceleration situation are the following: • Should a deceleration roadway be modeled by a series of decreasing-speed roadway segments, or should it be mod- eled by a constant-speed segment? • If using decreasing-speed roadway segments, over what distance should deceleration be modeled, and what should the speeds be? • Once vehicles have decelerated to a stop at a signal, should there be some accounting of the time spent stopped and Figure 6. ZOIs for a deceleration roadway as defined in NCHRP Report 311. 9 Menge, C. W., C. F. Rossano, G. S. Anderson, and C. J. Bajdek, FHWA Traffic Noise Model®, Version 1.0—Technical Manual, U.S. Department of Transporta- tion, Research and Special Programs Administration, John A. Volpe National Transportation Systems Center, Report No. DOT-VNTSC-FHWA-98-2 and FHWA-PD-96-010, Cambridge, MA, February 1998. 10 Bowlby, W., R. L. Wayson, and R. E. Stammer, Jr., NCHRP Report 311: Predict- ing Stop-and-Go Traffic Noise Levels, Transportation Research Board, National Research Council, Washington, D.C., 1989.

14 idling while the signal is red before acceleration begins when the signal turns green? • When is this entire effort worth doing? In other words: – For a one-way road, what is the effect of an acceleration roadway starting at the stop line on levels for receivers on the deceleration side of the stop line? – For a one-way road with a traffic signal, what is the effect of a percentage of the traffic on the deceleration side on the signal traveling at cruise speed? – For a two-way road with a traffic signal, what is the effect of the accelerating and cruising traffic on the far road- way on the levels for receivers along the deceleration side of the near roadway? 3.3 Research Tasks The research began with a survey of practitioners. Infor- mation was obtained on the modeling approaches used on a number of previous highway project noise studies, including several that were conducted by members of the research team. Following an analysis of previous modeling approaches, a sensitivity analysis was conducted to examine the size of the effects on the wayside Leq caused by varying the different input parameters for acceleration roadways. A single TNM inter- rupted flow roadway was created with an array of receivers along its length and offset to the side. Separate runs were made for automobiles and heavy trucks. Multiple runs were made for varying vehicle speeds (the final cruise speed after accel- eration), Speed Constraint, and the percentage of Vehicles Affected. Constant speed runs were also made for compari- son to the acceleration roadway runs. The next step was to address deceleration, initially through a sensitivity analysis. Test scenarios were developed using the deceleration roadway segment lengths and equivalent speeds for automobiles, medium trucks, and heavy trucks developed in the NCHRP Report 311 research. These scenarios were for the different approach speeds used in NCHRP Report 311 and, again, for an array of receivers. Then, scenarios of decreasing speed were developed based on the modeling approaches presented in the reviewed studies—an incremental decreasing of speeds from one seg- ment to the next. In addition, the scenarios were then re-run without any deceleration modeling at all to determine the effect, if any, of leaving out deceleration completely. Then, more realistic situations for interchanges were exam- ined such as (1) a diamond interchange ramp in the presence of a mainline highway and (2) an SPUI. For the diamond interchange, a scenario was created with a mainline and exit ramp and an array of receivers to test the effects of vari- ous multipliers of the ramp traffic volume on the mainline roadways. The goal was to gain insight on when the main- line traffic dominates the received levels. For the SPUI, cases of detailed and simplified modeling were tested and compared. The research included limited field measurements and model validation along the exit ramp of a diamond interchange to refine the findings from the sensitivity testing. Some individual vehicle pass-by sound exposure levels (SELs) were also mea- sured for prediction of an SEL-based Leq(h) for comparisons to the TNM modeling results. Much of the sensitivity analysis discussed for the signal- ized interchange also applies to regular intersections. The sensitivity analysis was expanded to include the effect of add- ing traffic moving in the opposite direction—acceleration on a roadway in one direction alongside deceleration on the roadway in the opposite direction. A second consideration was the presence of cruising traffic in addition to the accel- erating and decelerating traffic for signalized intersections where a certain percentage of the traffic cruises through the intersection on a green signal. The third consideration was the needed level of detail for the modeling. The initial work done for interchanges and intersections was also applied to roundabouts. In particular, the sensitivity analysis provided important information on the variation in Deceleration Range (mph) Length (ft) Speed ZOI(1) (mph) Speed ZOI(2) (mph) Sinitial Sfinal ZOI(1) ZOI(2) Automobiles MT HT Automobiles MT HT 30 0 150 100 29 26 24 18 13 10 40 0 275 100 34 30 28 18 13 10 50 0 400 100 38 34 31 18 13 10 60 0 500 100 41 36 33 18 13 10 MT = Medium trucks. HT = Heavy trucks. Table 3. NCHRP Report 311 guidelines for modeling deceleration roadways with a final speed of 0 mph.11 11 Bowlby, W., R. L. Wayson, and R. E. Stammer, Jr., NCHRP Report 311: Predict- ing Stop-and-Go Traffic Noise Levels, Transportation Research Board, National Research Council, Washington, D.C., 1989. Table 7 (adapted), p. 32.

15 sound levels that can be expected with changes in the FHWA TNM Flow Control input parameters for acceleration and the NCHRP Report 311 guidelines for deceleration. From these results, initial guidelines were developed on the modeling of components of roundabout approach and departure legs and inner circulatory roadway. Limited field validation noise measurements were also made at a one-lane roundabout site. The site was modeled in FHWA TNM (and used in the above-described sensitivity testing) with the traffic counted during the measurements, allowing comparison of the measured and predicted levels. Some deceleration and acceleration individual vehicle pass- by SELs were also measured for comparison to the measured and modeled Leq(h). 3.4 Outcomes of the Research—Best Practices and How to Implement Them for a Noise Study or TNM Model 3.4.1 Signalized Interchanges—Diamond There are two main components to the diamond inter- change: the entrance ramps and the exit ramps. Modeling of the crossing road can also be important. 3.4.1.1 Entrance Ramp The ramp should be modeled as a flow control acceleration roadway that starts at the beginning of the ramp, with 100% Vehicles Affected. The Speed Constraint should be 10 mph, based on NCHRP Report 311. If the ramp carries more than 3% heavy trucks, the Speed Constraint could be increased to 15 or 20 mph because automobiles can make the turn onto the ramp at a higher speed before beginning the acceleration along the ramp. Figure 7 shows the predicted Leq(h) for separate runs of 1,000 automobiles/hr and 1,000 heavy trucks/hr accelerating up to a cruise speed of 60 mph for Speed Constraints of 0 and 10 mph. The receivers are along the side of the roadway, offset 100 ft from it, beginning at the start line and proceed- ing downstream. These runs show the effect of the 10-mph Speed Constraint over the first several hundred feet of accel- eration and also show the large difference in Leq(h) by vehi- cle type if the volumes of the two vehicle types are equal. If the heavy truck percentage was only 3% of the total volume, the automobile and heavy truck curves would be roughly equal at the close-in distance and the automobile Leq(h) would dominate the total Leq(h) beyond about 1,000 ft downstream. Figure 8 shows this case for a Speed Constraint of 0 mph. The range in predicted Leq(h) over the entire Figure 7. Leq(h) for 1,000 automobiles and 1,000 heavy trucks, plotted separately, accelerating from 0 mph to 60 mph with 100% of the Vehicles Affected and Speed Constraint of 0 and 10 mph for a series of receivers offset 100 ft from the roadway.

16 acceleration region and on into the cruise speed region is only about 2 dB. The FHWA TNM roadway segment lengths should not exceed 50 ft if the final cruise speed is 30 mph, 100 ft for 45 mph, and 500 ft for 60 mph or higher. In the process of doing the sensitivity analysis for accelera- tion, an apparent error in the FHWA TNM speed algorithm was found. The FHWA TNM Technical Manual indicates that speed along an interrupted flow roadway segment is com- puted on a subsegment basis as the program subdivides user- specified segments for its sound-level computations. Then, when the vehicle reaches the “target” or final speed that has been input by the user, FHWA TNM is supposed to stop accel- erating the vehicle, revert back to the cruise emission levels, and continue computing levels along the roadway at the tar- get speed. Instead, it was found that once TNM accelerates the vehicle to the target speed, it does not revert back to the cruise emission levels until the beginning of the next roadway segment. The result is that the user’s choice of the segment lengths can result in very different predicted sound levels at a receiver instead of being independent of the user-specified segment length. The problem was studied, and guidance to avoid incorrect sound-level calculations was developed. Appendix B documents the analysis (available on the NCHRP Project 25-34 web page at http://apps.trb.org/cmsfeed/TRB NetProjectDisplay.asp?ProjectID=2986.) There are several options on the modeled length of the ramp roadway. Two of these options are the following: • Option 1 is to model the roadway past the physical merge point and then parallel to and offset by a foot from the outer mainline roadway until the end of the modeled mainline roadway. One advantage of this option is that the analyst does not have to determine where to stop the ramp roadway in terms of its effects on the total wayside sound level. A second advantage is that the analyst does not have to adjust the mainline roadway traffic volumes past the merge point to add in the ramp traffic. • Option 2 is to end the ramp at the physical merge point. Because of the difference in the cruise and “full throttle” REMELs in FHWA TNM, unless the ramp truck traffic is a very large percentage of the mainline truck traffic (40% at 50 mph and 16% at 70 mph), the consequence of ending the ramp before the cruise speed is attained and modeling all of the trucks beyond that point at the cruise speed is slight (±0.5 dB) over-prediction of level (less than 0.5 dB). As a guide, it is sufficient to have FHWA TNM only accelerate the heavy trucks up to a speed of 30 mph—a distance of about Figure 8. Leq(h) for 97% automobiles and 3% heavy trucks (31,000 vehicles), accelerating from 0 mph to 60 mph with 100% of the Vehicles Affected and a Speed Constraint of 0 mph for a series of receivers offset 100 ft from the roadway.

17 700 ft on a 0% grade—for the above-cruise speeds and truck percentages. Appendix B provides details on this analysis. 3.4.1.2 Exit Ramp The need to model deceleration along the ramp in detail is moderated by several factors. First, while 100% of traffic will either have to stop at the signal or decelerate down to about 10 to 20 mph to make a turn at the end of the ramp, the traf- fic may then be modeled as accelerating away from the end of the ramp. If there is a queue on the ramp for the signal, that acceleration will occur along the ramp. Acceleration from the end of ramp or the queue will affect levels at upstream receivers; as a result, precise modeling of end of decelera- tion is not needed. Second, the mainline noise may be the dominant contributor to the total sound level for receivers along the ramp; the effect is a function of the receiver offset distance from the ramp, the distance upstream along the ramp, and the amount of traffic on the ramp compared to the main- line traffic. Figure 9 compares the wayside Leq(h) for receivers at an offset distance of 100 ft along the deceleration roadway for a deceleration roadway alone and the deceleration roadway with an acceleration roadway heading downstream from the stop line away from the upstream receivers. The stop line is on the left of the chart and upstream is to the right. Sepa- rate predictions are shown for 1,000 automobiles and 1,000 heavy trucks with 100% of the Vehicles Affected and a Speed Constraint of 0 mph for the acceleration. The deceleration roadway was modeled using the NCHRP Report 311 guide- lines for 60-to-0-mph deceleration. The effect of the down- stream acceleration away from the upstream receivers is large. The levels are higher than the deceleration-only case as far upstream as 300 ft from the stop line. In real-world terms, the acceleration of the vehicles away from the stop line is heard upstream at a level high enough to affect the total upstream level. The greater the offset distance, the greater the influence will be, as shown in Appendix B. When an acceleration roadway is modeled at the end of a deceleration roadway, precise modeling of at least the last 100 ft of the deceleration roadway is not needed—the total level will be largely influenced by the acceleration roadway. Field noise measurements along an exit ramp of a diamond interchange on Briley Parkway in Nashville, Tennessee, dem- onstrated this upstream effect from acceleration away from the stop line. Figure 10 shows the measured Leq(15 min) at a 50-ft offset from the ramp centerline at distances of 50, 100, 200, and 400 ft upstream from the stop line at the end of the ramp. The elevated measured Leq at the 50-ft point, and to some degree at the 100-ft point, show the effects of noise from traffic accelerating away from the ramp and passing local road traffic. Also shown are the FHWA TNM predictions with the Figure 9. Separate Leq(h) for 1,000 automobiles and 1,000 heavy trucks decelerating from 60 mph to 0 mph using NCHRP Report 311 roadway segment lengths and speeds and then accelerating downstream from 0 mph to 60 mph for a series of receivers offset 100 ft from the deceleration roadway.

18 ramp traffic counted during three of the four measurement periods factored up to hourly volumes. The deceleration was modeled using the NCHRP Report 311 segment lengths and speeds for deceleration from 60 to 0 mph acceleration at the end of the ramp, and local road traffic are not modeled. The model under-predicts the most at the 50- and 100-ft upstream points. Measurements of individual vehicle sound levels dur- ing deceleration along the ramp showed a trend similar to that of the modeled levels. Testing was then done to improve the prediction of the deceleration Leq by FHWA TNM in comparison to an Leq(h) computed based on the measured SEL data. The results for the exit ramp suggest that the NCHRP Report 311 segment lengths remain valid when modeling deceleration from 60 mph, but with revised speeds: • Roadway Segment ZOI(1): 500 ft long with speeds of 50, 40, and 35 mph for automobiles, medium trucks, and heavy trucks, respectively. • Roadway Segment ZOI(2): 100 ft long with a speed of 20 mph for each vehicle type. However, the results also show that the Leq at the 50-ft and 100-ft upstream distances from the stop line are heavily influ- enced by the noise of the vehicles accelerating away from the stop line. The influence of mainline traffic noise on levels for receiv- ers along a deceleration ramp can be seen in the results of some additional modeling at another diamond interchange. A receiver array was set up with offsets of 50, 100, 200, and 400 ft from the ramp and spacing upstream along the ramp at distances from 50 to 800 ft from the stop line. The end of the ramp is approximately 520 ft from the outermost mainline travel lane. Tests were made for several ratios of mainline traffic to ramp traffic: mainline traffic equal to 2, 4, 8, and 16 times the ramp traffic, translating to ramp traffic percentages of the mainline traffic of 50%, 25%, 12.5%, and 6.3%, respectively. Only automobiles and heavy trucks were mod- eled, with a mix of 9% heavy trucks on both the mainline and the ramp. Figure 11 shows an example of results for receivers close to the ramp (offset 50 ft from the ramp centerline) for a high percentage of ramp traffic—25% of the mainline traffic. In this case, the ramp traffic is an important contributor to the total Leq(h), especially at the shorter distances upstream. Figure 12 shows the results for receivers farther from the ramp (offset 200 ft) and for a lower ramp traffic percentage—12.5% of mainline). For these receivers, the ramp only affects the total Leq(h) by 0.5 dB or less regardless of the distance upstream from the ramp stop line. Not included in the results are the effects of acceleration away from the ramp or traffic on the cross street. If these conditions were modeled in the runs, the need for accurate modeling of the deceleration ramp would decline, even for receivers closer to the ramp and closer to the stop line. More results are available in Appendix B. 3.4.2 Signalized Interchanges— Folded Diamond The folded diamond has one pair of entrance and exit ramps in the traditional diamond layout and the other pair as loop ramps onto and off of the mainline, as shown in Figure 10. Comparison of measured traffic noise and modeled noise excluding mainline heavy trucks, Briley Parkway exit ramp site.

Figure 11. Diamond interchange Leq(h) for ramp traffic equal to 25% of mainline traffic using NCHRP Report 311 deceleration roadway segment lengths and speeds, for a series of receivers offset 50 ft from the ramp centerline. Figure 12. Diamond interchange Leq(h) for ramp traffic equal to 12.5% of mainline traffic using NCHRP Report 311 deceleration roadway segment lengths and speeds, for a series of receivers offset 200 ft from the ramp centerline.

20 Figure 13. Traffic signals control the flow on either side of the interchange. The end of the exit ramp and start of the entrance ramp at each signal resemble a two-way road intersection with approach and departure roadways on one leg only. 3.4.2.1 Entrance Loop Ramp The FHWA TNM roadway would start at the beginning of the ramp, just past the traffic signal. It would be designated as a flow control roadway with 100% Vehicles Affected and a Speed Constraint of 10 mph (based on NCHRP Report 311 for heavy trucks) until the loop curve is reached. Then, a new roadway of cruise segments would be used to model the loop at the posted ramp speed. Then, an acceleration road- way would be modeled with 100% Vehicles Affected and a Speed Constraint equal to the ramp loop speed up to final mainline speed. The FHWA TNM roadway segment lengths should not exceed 50 ft if the final cruise speed is 30 mph, 100 ft for 45 mph, and 500 ft for speeds of 60 mph or higher. 3.4.2.2 Entrance Diamond Ramp This ramp would be modeled in the same manner as described for the regular diamond interchange. 3.4.2.3 Exit Loop Ramp The FHWA TNM roadway would start at the beginning of the ramp. It would be modeled by a series of segments along the loop at the posted speed. A final 100-ft segment could be modeled at a speed of 20 mph, ending at the stop line. The Figure 13. Folded diamond interchange.12 immediately adjacent entrance ramp with accelerating traffic and the local crossing road would dominate the levels for any nearby receivers. 3.4.2.4 Exit Diamond Ramp This ramp would be modeled in the same manner as described for the regular diamond interchange. The immediately adjacent entrance ramp with accelerating traffic and the local crossing road would dominate the levels for any nearby receivers; precise modeling of the deceleration is much less important. 3.4.3 Signalized Interchanges—Single Point Urban Interchange SPUIs present potentially complex modeling scenarios. The mainline can be designed to pass over or under the turn- ing movements. When passing under, the mainline traffic is largely shielded from the receivers by the ramp embankments or retaining walls. When the mainline traffic passes over the crossing road, mainline noise will dominate the exit ramp traffic’s deceleration noise even more than at diamond inter- changes because the SPUI ramp is closer to the mainline due to geometry of the interchange design. Also, in this configura- tion, the interchange movements are under the mainline deck and are shielded from the receivers. 3.4.3.1 Full Modeling Full modeling is generally not necessary, but the details of the center intersection movements will be briefly described as a basis for understanding the partial modeling. Figure 14 shows the TNM plan view (assuming north is at the top of the figure) for full modeling of a SPUI studied in this research, where the mainline passes under the interchange deck. Fig- ure 15 shows a detail of the modeling of the top of the ramp deck. In the detail: • The thick solid line is the eastbound crossing road street section going across the deck, represented by a flow control roadway that accelerates traffic from the traffic signal. • The dotted line is a flow control roadway representing southbound exiting traffic that accelerates to the east away from the signal at the end of the ramp. • The dashed line is a flow control roadway representing entering traffic from the eastbound cross street accelerat- ing to the north away from the signal at the start of the ramp lane. The flow control parameters for each of the three indicated roadways are a speed constraint of 0 mph and an assumed 50% of the traffic affected by the signal. There are three 12 Imagery © 2014 Google, Map data © 2014 Google.

21 similar flow control roadways for traffic moving in the oppo- site direction on the crossing road. A disadvantage of this detailed modeling is that it requires modeling of 12 inter- secting points of the ramp sections and the cross street road- ways, all set as “on structure” segments to allow the mainline roadways to pass under them. 3.4.3.2 Partial Modeling Partial modeling of the interchange turning movements has the advantages of avoiding micro-modeling of all segments of all turning movements and avoiding modeling of 12 roadway segment intersecting points in the center deck area. A disad- vantage is that partial modeling may slightly underestimate sound levels for receivers very close to the end of the exit ramp because the acceleration away from the signal for the left leg of the ramp is not modeled. Figure 16 shows the TNM plan view for the partial model- ing. This partial modeling method requires only four flow con- trol roadways for accelerating traffic on the deck: eastbound crossing road (A), westbound crossing road (B), northbound entrance ramp (C), and southbound entrance ramp (D), and two more flow control roadways not on the deck—the eastbound-to-southbound entrance ramp (E) and the west- bound-to-northbound entrance ramp (F). Partial modeling Figure 14. FHWA TNM plan view for full modeling of an SPUI (north is to the top). Figure 15. Detail of the modeling of the top of the ramp deck for the full modeling case. also eliminates all of the otherwise needed roadway intersec- tion points (FHWA TNM requires that crossing roadways share a common point with identical x, y, and z coordinates). Entrance Ramps. There is one entrance ramp in each mainline direction—northbound (G) and southbound (H)— with each ramp consisting of two acceleration roadway sec- tions. The dashed line toward the center of Figure 16 and the interchange is the eastbound-to-northbound entrance ramp roadway (C). It is modeled as a flow control roadway starting at a point that is past the crossing points in the cen- ter of the deck; its branch on the right is the westbound-to- northbound entrance ramp roadway (F), which has no flow control device. The FHWA TNM roadway segment length for the flow control roadways will depend on the final desired cruise speed, as described for the diamond interchange. The right-turn (e.g., westbound-to-northbound) entrance ramp roadway (F, shown as a dashed line) starts at a point beyond the crossing points on the deck. It is modeled as a flow control acceleration roadway with 100% Vehicles Affected. The Speed Constraint should be 10 mph if this movement is a full right turn at a signal and can be 20 to 25 mph for chan- nelized flow that eliminates the full right turn.

22 For the left-turn movement (C), the flow control accel- eration roadway is started past the center of the interchange with 50% Vehicles Affected due to the presence of the signal, which is on the entrance (western) side of the deck and a Speed Constraint of 20 mph because the vehicles are already moving forward from the signal. The southbound entrance ramp (H) is modeled in the same way as the northbound entrance ramp (G). Both ramps are extended to their physical merge points with the main- line, which provides sufficient length for acceleration of the heavy trucks to minimize any under-prediction caused by not extending the ramp until the final cruise speed is actually reached. Exit Ramps. There is one exit ramp in each mainline direction—southbound (I) and northbound (J)—with each ramp consisting of two branches near its end. The south- bound exit ramp (I) is represented in the figure by dotted lines. The branch to the left (in the direction of travel) is the southbound-to-eastbound exit ramp roadway (K) and is modeled as ending at a traffic signal. The branch to the right (in the direction of travel) is the southbound-to-westbound exit ramp roadway (L), which may or may not end at a signal, depending on the design. Precise deceleration modeling is not critical because of the acceleration at the end of the ramp toward the outside of the interchange (L)—not the movement across the cen- ter of the interchange (K)—and acceleration of the adjacent crossing road’s through traffic (B). The exit ramps may be modeled as a series of segments with decreasing speeds. The mainline speed is carried well along the ramp. NCHRP Report 311 ZOI(1) and ZOI(2) seg- ment lengths may be used for the left branch (K) to the signal leading onto the center deck. The speeds for those segments can be the speeds derived from the diamond interchange noise measurements and modeling. For the branch to the right (L), the speed will depend on whether the branch is channelized for smooth merging into the crossing road or signalized. If channelized, speeds of 25 to 35 mph could be used depending on the geometrics. The right branch of the ramp (L) would end at the physical merge point with the crossing road (B). If signalized, the last segment (ZOI[2]) would be at 20 mph, in which case the crossing then starts as an acceleration roadway. Crossing Road. For the crossing road, each travel direc- tion should be modeled separately. The thick solid line in Figure 16 is the eastbound crossing roadway, broken into two modeling sections. The one on the left is the eastbound approach leg (M) and is modeled as ending at a traffic signal at the “entrance” to the deck. It may be modeled at cruise speed because of the acceleration on the nearby southbound entrance ramp legs (D and E) and acceleration in the opposite direction by the westbound crossing road’s departure traffic. The thick solid line on the right of the figure is the east- bound crossing roadway’s departure leg (A). It is modeled as starting past the center of the deck as a flow control road- way with 50% of Vehicles Affected and a Speed Constraint of 20 mph because vehicles have already been moving forward from the signal at the entrance to the center deck. 3.4.3.3 Discussion To illustrate the effect of the acceleration of far lane traf- fic in the opposite direction and adjacent to the decelerat- ing near lane, Figure 17 compares results for modeling the approach leg with decelerating vehicles using the NCHRP Report 311 recommendations against modeling the approach at a cruise speed of 30 mph and for heavy trucks only. The accelerating traffic in the other direction was modeled on a flow control roadway for Vehicles Affected values of 25%, 50%, and 75%. Both roadways represent single lanes sepa- rated by 12 ft. In this particular case, the stopping points in each direction were offset 80 ft to simulate their separation Figure 16. Detail of the SPUI modeling for the partial modeling case.

23 at a typical intersection of two two-lane roads. The separa- tion would be greater for the SPUI. In this situation, receivers on the upstream deceleration side of the near roadway are exposed to the noise of accelerating vehicles on the down- stream side of the acceleration roadway directly across from them. The results are for a string of receivers along the near road, offset from it by 50 ft. When comparing the cases for each percentage of Vehicles Affected, the modeling of the deceleration conditions on either side does little to reduce the total received level, due to the dominance of the acceleration roadways’ noise on both sides of the intersection. Modeling the approach roadway at the 30 mph cruise speed is sufficient. The key is to model the acceleration roadway in each direc- tion. As the receiver offset distance increases, the differences decrease, meaning that there are fewer cases in which it may be necessary to model deceleration in the presence of cruising traffic. More details on this analysis are in Appendix B. As a conservative worst case, modeling on the deceleration side could be done with all cruising traffic first, and if levels were within 1 dB of causing impacts, more precise modeling might be needed to demonstrate that no impacts are predicted to occur. The results of the comparisons of the full and partial mod- eling of a SPUI where the mainline passes under the cross street indicate the following: • Partial modeling of the interchange on the deceleration ramp side is sufficient if there are no receivers within 300 ft of the intersecting road or 400 ft of the deceleration ramp. Even then, detailed modeling is not needed if the partial modeling shows the levels are more than 1 dB below the noise impact criteria in the state highway agency noise policy. • Partial modeling of the interchange on the acceleration ramp side is sufficient even if there are receivers very close to the intersecting road or ramp. • When partially modeling the SPUI deck, the speed con- straint for both the cross street and the entrance ramp should be 20 mph, with 50% of the Vehicles Affected. For the entrance ramp not crossing the deck, the Speed Constraint should be based on the geometrics of the ramp (20 mph was used in the testing). Where the mainline passes over the cross street, the par- tial interchange modeling should be sufficient in almost all cases because of the shielding of the interchange movement beneath the mainline and the greater exposure of the receiv- ers to the mainline noise. Details of the testing of the various scenarios and the results are in Appendix B. 3.4.4 Intersections—Unsignalized 3.4.4.1 Two-Way Stop This situation would involve a more heavily traveled main road and a lower volume cross street. The main road should Figure 17. Leq(h) for 1,000 heavy trucks for two-way “Deceleration 1 Cruise” compared to “All Cruise” on the upstream deceleration side for a 30-mph cruise speed and an array of receivers at a 50-ft offset distance from the roadway.

24 be modeled by FHWA TNM roadways in each direction at cruise speed with no acceleration or deceleration. The cross street probably does not need to be modeled because if even a four-way stop is not warranted to control traffic on the main road, then intersecting road volumes and speed are both likely to be low. However, the cross street could be modeled if there are adjacent receivers by a flow control acceleration roadway starting just past the mainline road- ways using 100% Vehicles Affected and a Speed Constraint of 20 mph to represent speed as the vehicle exits the inter- section. The local road approach leg should be modeled at the posted speed for that road; no modeling of reduced speeds for deceleration is needed if the approach speeds are 40 mph or less. 3.4.4.2 Four-Way Stop The four-way stop may require more complete modeling if there are receivers adjacent to each road. One would model the acceleration away from the stop line in each of the four directions. Total modeling would require many intersecting points for the crossing roadways because FHWA TNM does not allow two roadways to cross without sharing a point with the same x, y, and z coordinates and may not be needed: • If the scenario is modeled with one FHWA TNM roadway in each direction, there would be four points of intersection. • If the scenario is modeled as two FHWA TNM roadways per direction of travel on one road and one FHWA TNM roadway per direction of travel on the other road, there would be eight intersecting points. • If the scenario is modeled as two FHWA TNM roadways in each direction for each road, there would be 16 intersecting points. In all cases, the flow control roadway would start at the stop line with 100% Vehicles Affected and a Speed Constraint of 0 mph. As illustrated in the SPUI discussion, the approach- ing traffic could be modeled as the posted speed. If the posted speed were as high as 60 mph, there would be over-prediction by 1–3 dB by not modeling the deceleration. A simpler approach is to partially model the movements of one of the roads and avoid all of the intersecting FHWA TNM points. In this case, model the road with the most traf- fic (or perhaps the most adjacent receivers) as continuous, with an FHWA TNM cruise speed roadway on the upstream side connected at the stop line to a flow control acceleration roadway that crosses through the intersection and proceeds downstream on the departing leg. The flow control roadway would have a Speed Constraint of 0 mph and 100% Vehicles Affected. The lesser road would be modeled as described above for the two-way stop: (1) on the departing leg, by a flow control acceleration roadway starting just past the main roadways using 100% Vehicles Affected and a Speed Constraint of 20 mph to represent speed as the vehicles exit the intersection, (2) on the approach leg, by a constant-speed roadway at the posted speed for that road. No modeling of reduced speeds for deceleration is needed unless the posted speed is high and the simpler modeling did not result in levels within a couple of dB of causing noise impacts. 3.4.5 Signalized Intersections 3.4.5.1 One-Way Roadways Model the departing leg as a flow control acceleration roadway starting halfway back along the upstream queue. Use 50% Vehicles Affected, a Speed Constraint of 0 mph, and a final speed of the operating or posted speed. Model the approaching leg as a constant-speed roadway at the operating or posted speed to halfway back in the queue. The low-speed deceleration does need to be modeled unless the posted speed is high because of the dominance of noise from the percentage of traffic cruising through the signal and the per- centage of traffic accelerating from a stopped condition on the upstream side of the intersection. The effect of the accelerating traffic on the upstream receiver levels was shown in the discussion on diamond interchanges. The effect of two-way traffic with accelera- tion in each direction was shown in the SPUI discussion. Figure 18 and Figure 19, shown here, are similar to what was shown for the two-way road, except that they represent a single road. Cruise speeds of 30 and 60 mph were tested for Vehicles Affected values of 25%, 50%, and 75% for the accel- eration roadway, with those same percentages applied to the deceleration side of the intersection. Because prior analysis showed the dominance of heavy truck noise over automobile noise except at very high percentages of automobiles, only heavy truck cases were run. Figure 18 illustrates the results for 1,000 heavy trucks for a cruise speed of 30 mph and at a receiver offset distance of 50 ft. For all three percentages of Vehicles Affected, there is very little difference in the levels for the “Deceleration + Cruise” case compared to the “All Cruise on the Decelera- tion Side” case. At a low speed, the combined presence of the cruise traffic on the upstream deceleration side and accelerat- ing traffic on the downstream acceleration side dominates the total level for upstream receivers. Figure 19 shows the results for the cruise speed of 60 mph. The levels on the upstream side of the intersection for the “Deceleration + Cruise” are lower than the levels for the “All Cruise on the Deceleration Side” cases from the

25 Figure 18. Leq(h) for 1,000 heavy trucks for one-way “Deceleration  Cruise” compared to “All Cruise” on the upstream deceleration side for a 30-mph cruise speed and an array of receivers at a 50-ft offset distance from the roadway. Figure 19. Leq(h) for 1,000 heavy trucks for one-way “Deceleration  Cruise” compared to “All Cruise” on the upstream deceleration side for a 60-mph cruise speed and an array of receivers at a 50-ft offset distance from the roadway.

26 stopping point back to an upstream distance of 600 ft by the following: • 25% Affected: 0.4 to 0.8 dB. • 50% Affected: 0.6 to 1.6 dB. • 75% Affected: 0.9 to 2.6 dB. As the offset distance to the receivers increases beyond 50 ft, the differences decrease (as shown in Appendix B), meaning there are fewer cases in which it may be necessary to model deceleration in the presence of cruising traffic. Thus, for a 30-mph case, the approach side of the intersection may be modeled by a cruise roadway at the desired speed. For the higher speed, as a conservative approach, the deceleration side may initially be modeled by a cruise roadway at the cruise speed. If the predicted Leq(h) is high enough to cause noise impacts, then more detailed modeling of the deceleration may be needed to confirm the existence of impacts. 3.4.5.2 Two-Way Roadways The degree to which a signalized intersection with two- way traffic on all legs needs to be modeled depends on the proximity of the receivers. As described for the four-way stop, total modeling would require many intersecting points for the crossing roadways because FHWA TNM does not allow two roadways to cross without sharing a point with the same x, y, and z coordinates and may not be needed. A simpler approach is similar to what was described for the four-way stop—partially model the movements of one of the roads and avoid all of the intersecting FHWA TNM points. The road with the most traffic (or perhaps the most adjacent receivers) would be modeled as continuous in each direction. A constant speed FHWA TNM roadway (or multiple road- ways for multiple lanes) would be modeled on the approach, connected to a flow control acceleration roadway (or road- ways) that crosses through the intersection and proceeds downstream on the departing leg. The joining point would be halfway up the expected queue, which could be several hun- dred feet from the stop line. The flow control roadway would have a Speed Constraint of 0 mph and 50% Vehicles Affected. The intersecting road would be modeled as not crossing through the intersection. On the departing leg, a flow control acceleration roadway would start just past the main roadways to avoid the intersecting points. This flow control roadway would have 50% Vehicles Affected and a Speed Constraint of 20 mph to represent the speed as the vehicles exit the intersection. On the approach leg, a constant-speed roadway would be modeled at the posted speed for that road; no modeling of reduced speeds for deceleration is needed. Unlike the one- way road case illustrated above, even at higher approach speeds, the difference between modeling a combination of deceleration and cruise and all cruise is small. Figure 20 shows the results for the two-way situation at 60 mph at a receiver offset distance of 50 ft. As with the Figure 20. Leq(h) for 1,000 heavy trucks for two-way “Deceleration  Cruise” compared to “All Cruise” on the upstream deceleration side for a 60-mph cruise speed and an array of receivers at a 50-ft offset distance from the roadway.

27 30-mph cases, there is symmetry on the approach and depar- ture legs caused by the accelerating traffic in each direction. The largest differences at the 50-ft receiver offset distance are on the order of 1 dB. As shown in Appendix B, as the receiver offset distance increases, the differences decrease, meaning that there are fewer cases in which it may be necessary to model deceleration in the presence of cruising traffic. As a conservative worst case, modeling on the decelera- tion side could be done with all cruising traffic first and, if levels were within 1 dB of causing impacts, the more precise modeling might be needed to demonstrate that no impacts are predicted to occur. 3.4.6 Roundabouts Roundabout design is largely governed by guidance in NCHRP Report 672. Key design factors are the entry, circu- lation, and exit speeds, which are determined by the radii of the curves leading into, going around, and leaving the round- about. For a single-lane roundabout with a center radius on the order of 90 ft or less, the typical entry and circulation speed is 20 mph. For a multilane roundabout with a larger radius, the typical entry and circulation speed is approximately 25 mph. This research focused on the slower-speed roundabout with a one-lane inner circulatory road, but tests were also made for higher speeds and have been generalized to the larger two-lane inner circulatory road. This research showed that detailed noise modeling of all of the roundabout movements is generally not needed. How- ever, if one chooses to model all of the movements, the meth- odology illustrated in Figure 21 for the eastbound “through” movement of the east-west roadway is a good way to repre- sent speeds and assign traffic volumes for the deceleration, constant-speed, and acceleration components. Figure 21 shows an approach FHWA TNM roadway (thick solid line) that models (1) constant speed to the beginning of deceleration (not shown); (2) deceleration with two decreasing speed segments (only ZOI[2] is shown); (3) a final constant- speed approach segment (“Entry”); (4) two segments repre- senting the east-west portion of the circle (which, as described below, do not necessarily have to be modeled with traffic on them); and (5) a constant-speed “Exit” segment with the circle speed out to where acceleration away from the round- Figure 21. Modeling methodology for eastbound approach and departure legs and center circle of roundabout.

28 about would begin. Figure 21 also shows a departure FHWA TNM roadway (dotted line), modeling acceleration with a flow control device of “Onramp,” a Speed Constraint equal to the speed on the last segment of the approach roadway, and a final desired cruise speed (in these runs equal to the approach speed). 3.4.6.1 One-Lane Inner Circulatory Road Approach Leg. The approach to the roundabout may be modeled by a constant speed equal to the posted speed up to the beginning of the splitter island/crosswalk. Then, one 25-mph segment would be used to represent the entry leg, ending at the entry point to the circulatory road. Inner Circulatory Road. The traffic on the inner circula- tory road does not need to be modeled. The noise from the accelerating traffic departing the roundabout will dominate the overall sound levels. Departure Leg. For the departure leg, a one-segment constant-speed roadway would be modeled at a speed of 25 mph. It would start at the exit point from the inner cir- culatory road and end at the end of the reverse curve typi- cally at the end of the splitter island/crosswalk. Then, a flow control acceleration roadway would be modeled from the point downstream to the end of the modeled site. The road- way would have a Speed Constraint of 25 mph and 100% Vehicles Affected with the posted or operating speed as the final desired speed. 3.4.6.2 Discussion Sensitivity testing was based on an actual single-lane roundabout that was modeled in FHWA TNM and was also studied in the field. The location was the western side of the Liberty Pike roundabout at Turning Wheel Lane in Franklin, Tennessee. Most of the runs were made for cruise speeds of 40 mph. Additional runs were also made for 30 and 50 mph. Because most roundabouts appear to carry predominantly automobile traffic, the runs were made for automobiles only. The analysis showed that the accelerating noise of traffic on departure from the roundabout dominated sound levels close in. As a result, little, if any, change was seen in leaving the traffic off of, or not modeling, the inner circulatory road. Figure 22 presents a sample of the results for 40 mph for the 50-ft receiver offset distance, comparing two cases—Case 1 and Case 9: 65 66 67 68 69 -50 -100 -150 -200 -250 -300 -400 -500 -600 -700 -800 Le q( h) , d BA Distance from end of roadway,  Case 1 Approach Case 1 Departure Case 9 Approach Case 9 Departure Figure 22. Comparison of Leq(h) for Cases 1 and 9 for a series of receivers offset 50 ft from the western leg roadways for 1,000 automobiles approaching and departing roundabout at 40 mph.

29 • Case 1 included full modeling of all four legs of the round- about using NCHRP Report 311 deceleration segment lengths and speeds for 40 to 0 mph on the approaches, a circulatory road speed of 15 mph, and FHWA TNM acceleration on the departures. • Case 9 included the same deceleration modeling as Case 1 for the western, northern, and southern legs, but no traffic on the eastern legs or the circulatory road. Considering Case 1 or Case 9 individually, the levels on the (westbound) departure side are about 1 dB higher than on the (eastbound) approach side because those receivers are closer to the louder accelerating traffic. In comparing Case 9 to Case 1, not modeling traffic on the eastern leg and the circulatory road affected levels by only fractions of 1 dB, even at the closest receiver, located 50 ft from the entry/exit points in the circle. Figure 23 compares the same two modeling cases at 40 mph for a receiver offset distance of 200 ft. Again, there is very little difference between the Case 1 and Case 9 results along both the eastbound approach and westbound departure legs. The reason the levels are high at the points closest to the roundabout for the 200-ft offset distance is that these points are actually much closer to the modeled northbound and southbound approach and departure legs. For the 200-ft receiver offset distance along the western leg of the roundabout, the traffic on the circulatory road and on the eastern leg does not need to be modeled. Similar results were found for 30- and 50-mph scenarios. For both speeds, there is very little difference in the two cases, meaning the circulatory road and the eastern leg roadways do not need to be modeled. Thirty-minute measurements of the A-weighted Leq and individual vehicle SELs were made at four points along each western leg (approach and departure) of this roundabout at 50, 100, 200, and 400 ft from the entry/exit points and an offset distance of 25 ft from the edge of the travel lane. Traffic was almost entirely automobiles. Figure 24 shows the measured Leq(30 min) at each point for the approach-side measurement and the departure-side measurement. Also shown are the measured average SELs for automobiles for deceleration and for acceleration. The deceleration SEL values decrease by about 6 dB going from 400 ft to 50 ft upstream. In contrast, the acceleration SEL on the departure leg only increases by about 2 dB going from 50 ft downstream to 200 ft downstream, and then decreases 1 dB from 200 ft to 400 ft. The Leq data along each side tend to match the pattern of the acceleration SEL data—there is a relatively small increase from 50 ft to 400 ft. 54 55 56 57 58 59 60 61 62 63 64 65 66 -50 -100 -150 -200 -250 -300 -400 -500 -600 -700 -800 Le q( h) , d BA Distance from end of roadway,  Case 1 Approach Case 1 Departure Case 9 Approach Case 9 Departure Figure 23. Comparison of Leq(h) for Cases 1 and 9 for a series of receivers offset 200 ft from the western leg roadways for 1,000 automobiles approaching and departing roundabout at 40 mph.

30 Essentially, the noise of accelerating traffic on the departure side of the roundabout dominates the measured Leq on both sides of the road, supporting the sensitivity analysis conclusion. 3.4.6.3 Two-Lane Inner Circulatory Road A roundabout with a two-lane inner circulatory road may be modeled in essentially the same way as a one-lane inner circulatory road. 55 57 59 61 63 65 67 69 71 73 75 50 100 200 400 So un d Le ve l, dB A Distance from entry/exit point,  Approach Leq(30 min), dBA Departure Leq(30 min), dBA Deceleraon Auto SEL, dB Acceleraon Auto SEL, dB Figure 24. Measured Leq (30 min) at 25-ft offset from edge of travel lane, Liberty Pike roundabout. Because of the slightly higher speed typical of the two-lane case (20 to 25 mph instead of 15 to 20 mph on the smaller diameter one-lane road and the greater circumference), there might be a desire to model the inner circulatory road, espe- cially if receivers are immediately adjacent. However, if the inner road’s entry and approach legs are each modeled, then it is unlikely that the inner road itself needs to be modeled, especially because of the noise of vehicles accelerating away from the roundabout.