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.

45 C H A P T E R 6 6.1 Introduction Based upon the extensive FHWA TNM modeling experience of research team members and the review of data obtained from the literature search, a number of candidate modeling techniques for multilane highways have been identified. Typical issues encountered in FHWA TNM modeling of roadway sections that contain more than one travel lane in each direction include the following: • Modeling groups of lanes versus modeling each lane as its own roadway. • How much to overlap lanes. • How to represent shoulders and median areas. • How to represent edge of roadway section diffraction points. • Shielding of one roadway by another roadway, such as with a bifurcated roadway section. • Modeling super-elevated roadways. • It is envisioned that FHWA TNM Version 3.0 will be capa- ble of modeling multilane highways via its multilane tool; however, this version is not yet available for use and its limitations and graphic functionality are still being evalu- ated. Therefore, evaluations using FHWA TNM Version 3.0 were not conducted. In evaluating modeling techniques related to multilane high- ways, the research team focused on the bulleted items listed above, addressing traffic and noise measurements associated with locations with relatively simple topography. The evalua- tion and testing reinforced the team’s knowledge that factors such as pavement type, ground type, topography, noise barriers, and so forth affect noise levels. The influences of these factors were determined to often be more significant than the varia- tions of noise levels associated with the different techniques for modeling roadway lanes, shoulders, and median areas. While various pavement and ground types exist for the selected sites, no attempts were made as part of this investigation to address their variability in developing best modeling practices for multi- lane highways. In selecting measurement and validation data related to multilane highways, the team focused on collecting informa- tion for receptors located on adjacent land at elevations level with the highway, above the highway, and below the highway. Sites without median barriers were also selected, to elimi- nate this variable from the other variables examined. No data could be obtained for bifurcated highway projects without median barriers or outside parapets. Note that the topic of median barriers is addressed in Chapter 5. The techniques associated with FHWA TNM Version 2.5 were evaluated and tested using measurement data from the six selected projects described in Section 6.2 of this report. Sug- gested best modeling practices for modeling multilane high- ways were developed from this evaluation and testing. More detailed information is contained in Appendix E, which is available on the NCHRP Project 25-34 web page at http://apps. trb.org/cmsfeed/TRBNetProjectDisplay.asp?ProjectID=2986. 6.2 Measurement Locations Evaluated By far, the highest quality measurement and validation data exist in recent studies conducted by the Volpe Cen- ter. The following three locations in the Volpe measure- ment studies were selected for the evaluation of multilane highways: • Volpe Site AZ3B. This site is located in Arizona and is part of Volpe’s Arizona Quiet Pavement Program evaluation project. The site is relatively flat and the six-lane divided highway (three lanes in each direction) is relatively level. In 2008, three sets of measurements were taken at three primary locations—50, 95, and 246 ft from the center of the near lane. The reference (50-ft) microphone was positioned 5 ft above the roadway elevation, while the Multilane Highways

46 microphone heights at the 95-ft and 246-ft locations were positioned at 5 ft above the ground. • Volpe Site 01MA. In 2008, measurements were taken at this site, located adjacent to Route 24 in Massachusetts for use by Volpe in its FHWA TNM Validation Project. The site is relatively flat and the four-lane divided highway (two lanes in each direction) is relatively level. The paved shoul- ders were modeled as 10-ft wide roadways with no traffic. Measurements were obtained at 5 ft and 15 ft above the ground at distances of 50, 100, and 200 ft from the center of the near lane. • Volpe Site 20PA. This site is located adjacent to Inter- state 81 (I-81) west of Harrisburg, Pennsylvania, and is part of Volpe’s FHWA TNM Phase 2 Validation Project. In 2001, noise measurements were taken at four sites adjacent to the four-lane divided highway (two lanes in each direction) with a wide grass median. The sites were located at 90, 200, 400, and 600 ft on generally level terrain, except for the 600-ft site, which was approximately 14 ft higher than the others. • Volpe Site 19PA. This site is located adjacent to US Route 30 in Coatesville, Pennsylvania, and is part of Volpe’s FHWA TNM Phase 2 Validation Project. In 2001, noise measure- ments were taken at seven sites adjacent to the four-lane divided highway (two lanes in each direction) with a grass median. The sites were located at 50, 200, 400, 500, and 700 ft along a center offset row of microphones. In addition to the evaluation and testing of the noise mea- surement and traffic data available from the Volpe studies, the research team also evaluated and tested modeling techniques using four data sets from the following project, which used a simplified technique in modeling a multilane highway: • U.S. Route 35 Noise Analysis Project, Dayton, Ohio. In 2005, research team members conducted a variety of noise measurements for a section of US 35 as part of a preliminary noise evaluation for a proposed reconstruc- tion and widening project. Numerous measurement sites were located along this four-lane highway (two lanes in each direction) at locations level with the highway and above and below the elevation of the highway. Lanes were grouped and represented by a single roadway in each direc- tion, with the edge of shoulder diffraction edge defined by the outside edge of the modeled roadway closest to the measurement sites. In addition to evaluating measurements obtained at the above locations, the team considered measurements taken at an additional location adjacent to Interstate 95 in Philadelphia, Pennsylvania, in a limited evaluation of a multilane highway section where each roadway lane’s profile was independently modeled. Receptors at this location were located approxi- mately 50 and 100 ft from the highway and approximately 15 ft below roadway grade. 6.3 Evaluation of Modeling Techniques Based on a review of data collected by the team and input from team members and other noise specialists contacted, candidate modeling techniques were previously identified by the team. For evaluation and testing of each of these tech- niques, the team used the measurement and traffic informa- tion from the projects listed and described in Section 6.2. This process resulted in the development of best modeling practices to apply when modeling multilane highways using the FHWA TNM. Receptors located at various distances from the highway were evaluated using each of the techniques discussed below. In its evaluation of each of these modeling techniques, the team used 67 individual measurements and related traffic information associated with the five projects listed in Sec- tion 6.2. These measurements were taken at distances rang- ing from 50 to 700 ft from the center of the near traffic lane at points where the measurement site (microphone) ranged from approximately 20 ft below the elevation of the highway to approximately 29 ft above the highway. 6.3.1 Description of Modeling Techniques Candidate modeling techniques have been selected for basic FHWA TNM input elements related to roadways, shoulders, and diffraction edges, with consideration given to the bulleted issues listed in Section 6.1. For all projects, the ground type for any area existing between the inside shoulders was defined by the default ground type designated in the project’s FHWA TNM run. The three basic modeling techniques are described below. 6.3.1.1 Dummy Lane Technique This technique involves representing a shoulder in FHWA TNM by entering it as a roadway with a defined width and elevation and no traffic. The width of any designated outside dummy lane is typically set so as to also define the roadway section’s diffraction point. 6.3.1.2 Ground Zone Technique This technique involves defining a shoulder in FHWA TNM by representing the area of the shoulder with a ground zone. When representing a shoulder with a ground zone, the outside edge of the shoulder must be defined by a terrain line unless its elevation is the same as the adjacent topography.

47 6.3.1.3 Adjacent Lane Width Technique This technique involves defining a shoulder in FHWA TNM by establishing the outside of the shoulder by designating an appropriate width for its adjacent roadway lane. This width can also be used to define the outside diffraction edge of the roadway section. 6.3.2 Application of Techniques to Projects Various technique subcategories were also evaluated. They included modeling roadways (grouped lanes, individual lanes, and four options for overlapping lanes); modeling shoulders (dummy lanes, ground zones, adjacent lane width methods); and establishing roadway section diffraction edges (dummy lane, ground zone, and adjacent lane width methods). 6.3.3 Comparison of Modeling Techniques for Selected Projects Results of the comparison of the various modeling tech- niques addressed the difference between the measured noise levels and modeled noise levels for each of the three primary modeling techniques previously described—dummy lane, ground zone, and adjacent lane width. For each of these tech- niques, values are provided for four options for overlapping lanes plus a grouped-lane option. In applying the various modeling techniques to the proj- ects listed in Section 6.2, several conclusions were drawn. These conclusions are listed and discussed below and relate to the evaluation of these specific projects. • While there were a few outlier values, the vast majority of analysis sites showed little variation between the tech- niques in terms of the difference between measured values and modeled values. The average measured versus mod- eled absolute differences for each technique were approxi- mately 0.6 dB. • Even including the outlier values, the average measured versus modeled absolute differences for all techniques were each approximately 0.8 dB. The data suggested that the measured-modeled differences for sites located significantly lower than the elevation of the highway could be greater for techniques employing grouped roadway lanes versus those modeling individual lanes. However, sufficient receptors did not exist in the selected projects to verify this possibility. • Evaluation of the best modeling techniques in terms of dif- ferences between modeled and measured values suggested that it may be best to keep lane overlap distances in the 0.1-to-1.0-ft range and that using the dummy lane tech- nique may be the best. The ground zone technique employ- ing grouped lanes gave similar results, however. Most of the best results for the ground zone technique were associated with sites that were elevated with respect to the roadway. However, these ground zone trends are not sufficient to formulate a best modeling practice. For the projects described in Section 6.2, factors such as pavement type, ground type, distribution of traffic between lanes, measurement period variations, and vehicle speed iden- tification methods have the potential to create greater variation between measured and modeled values than do the different multilane modeling techniques evaluated. This potential, plus the fact that only a few insights into the development of a best modeling practice could be gleaned from the evaluation of the selected sites, prompted the team to consider a generic site where most of these factors could be normalized and where differences associated with the analysis techniques could be better determined. 6.3.4 Comparison of Modeling Techniques for a Generic Project In the construction of a project-validated model, the spe- cific effect of any individual input factor is not usually evi- dent. This was true in the team’s evaluation of the selected projects. For that reason, a generic project was developed and analyzed in an attempt to isolate the relative influences and differences between the multilane modeling techniques. The generic project considered a 4,000-ft-long, four-lane, divided highway with level grade and containing 10-ft-wide, paved, inside and outside shoulders and a paved median. Receptors were placed at setback distances of 50, 100, 200, 300, 400, and 500 ft from the center of the near lane at heights related to the highway of -15 ft, -5 ft, +5 ft, and +15 ft. For each of the primary modeling techniques, grouped lanes were modeled and compared to the use of individual lanes with 0.1-ft overlaps. The evaluation resulted in insignificant differences between individual- and grouped-lane modeling techniques with the four-lane generic project. This confirmed the general find- ings from the evaluation of the selected projects previously discussed. This finding was further validated by evaluation of the I-95 GIR project in Philadelphia. That project has super-elevated roadway lanes, but showed no significant dif- ferences between the individual- and grouped-lane modeling techniques. The results of the four-lane generic project evaluation led to the development and evaluation of a wider, eight-lane generic project that included additional receptors at eleva- tions of 25 and 35 ft below roadway grades. The results of the eight-lane evaluation showed that the grouped-lane tech- nique under-predicted noise levels relative to the individual- lane technique at receptors located close to and significantly lower than the highway. Presuming that individual-lane

48 modeling is more precise because the noise sources are more precisely located, this result illustrates the importance of modeling individual lanes in areas where certain lanes may be shielded and others may be exposed, or where certain vehicles in certain lanes are shielded and some are not. While this situ- ation most often exists in locations close to and/or below the grade of the highway, it could also exist at other locations that may be shielded or partially shielded by features that are either manmade (structures, barriers, etc.) or natural (undulating terrain, natural berms, etc.). To gather additional data related to the causes of the grouped- lane under-prediction of noise levels for the eight-lane generic project, FHWA TNM was run individually for automobiles, medium trucks, and heavy trucks, with results illustrating the predominance of the heavy truck noise component. This indicated that truck stack noise is a major component and a factor that must be considered in modeling roadway travel lanes in multilane highway situations. This will be of par- ticular significance for roadways with high truck volumes because trucks are predominant in the outside lanes, where they are close to the nearest receptors and will be less shielded by the edge of pavement where the roadway is elevated. 6.4 Best Modeling Practices Based on the evaluation of the analysis techniques reported, the research team has compiled a list of suggestions for mod- eling of multilane highway projects. This list represents the team’s best managing practices. The following two suggestions are deemed to be most important: • Model each travel lane separately when receptors are located below the elevation of the highway. • Regardless of the receptor’s relationship to the highway, model each travel lane separately when there are any inter- vening manmade or natural features that block the line of sight between any receptor and any travel lane. Consider roadway super-elevation and all perpendicular and flanking noise paths in making such determinations. If in doubt, model individual lanes. The following modeling techniques are suggested by the research team based upon the reported evaluations: • Set FHWA TNM default ground type to “Pavement” to minimize any possible effects created by inadvertently leaving gaps between roadways when modeling complex roadways with features such as ramp gores, curved road- way sections, and super-elevated roadways. Model median areas between paved shoulders and surfaces outside of the roadway section by use of the appropriate FHWA TNM ground zone(s). • Provide travel lane overlap distances in the 0.1-to-1.0-ft range. • Use the dummy lane technique to model shoulders, especially outside shoulders. It presents less potential for illegal intercepts within FHWA TNM and does not require the addition of a contour line that is required with the ground zone technique. The dummy lane tech- nique also allows for a smaller lane overlap than that resulting from use of the adjacent lane width technique and is more compatible with modeling super-elevated roadway sections. • When modeling super-elevated roadways, model the pro- file elevations associated with each roadway lane if such data are available.