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.
Page 3 2. Field Test Procedures This chapter describes the steps (tasks) involved in the field testing procedures employed in Phase 3 of the NCHRP 3-70 project. Task 0. Amplified Work Plan The objective of this task was to provide a detailed expansion of the approved research plan, along with refinements to the budget and schedule in response to Senior Program Officer and Panel comments. The Amplified Work Plan was submitted April 22, 2008. Panel comments were received June 3, 2008. The Revised Amplified Work Plan was submitted August 26, 2008. Task 1. Recruit Volunteer Agencies The objective of this task was to identify and recruit volunteer public agencies willing to contribute staff time learning, applying, and evaluating the NCHRP 3-70 Multimodal Level of Service (MMLOS) method. An adequate cross section of agencies operating urban streets in the United States was desired within the time and budget resources of the Phase 3 Continuation Project. The following volunteer agencies were selected and recruited to participate in the Phase 3 field tests: 1. Arlington County, Virginia 2. Atlanta Regional Commission (ARC), Georgia 3. City of Boise and ADA County Highway District, Boise, Idaho 4. Delaware Valley Regional Planning Commission (DVRPC), Philadelphia, Pennsylvania 5. City of Portland, City of Hillsboro, and City of Gresham, Oregon 6. San Antonio-Bexar County Metropolitan Planning Organization, San Antonio, Texas 7. City of San Diego, California The Florida Department of Transportation (FDOT), through a separate contract, also arranged for the participation of state highway planners and engineers, metropolitan planning organizations and cities in the following metropolitan areas: Tallahassee, Gainesville, and Tampa, Florida. These workshops and field tests were conducted by Scott Washburn of the University of Florida and his team, who prepared a report on the results of those workshops for FDOT. Task 2. Training Session, Selection of Field Test Arterials The objective of this task was to train the agency personnel on the NCHRP 3-70 MMLOS method and to select arterials for testing the method. Initial workshops and arterial selection were conducted at the locations and dates listed below. Summaries of the workshops, the arterial test results, and agency comments are provided in the attachments.
Page 4 Exhibit 1: Test Agencies Agency/Location Initial Workshop Arterials Tested Comments 1. Gainesville, Florida October 2, 2008 4 sponsored by FDOT 2. Tampa, Florida October 2, 2008 4 sponsored by FDOT 3. Tallahassee, Florida October 2, 2008 4 sponsored by FDOT 4. Orlando, Florida October 2, 2008 3 sponsored by FDOT 5. Philadelphia, Pennsylvania November 13, 2008 3 Hosted by DVRPC 6. San Diego, California December 10, 2008 2 Hosted by City of San Diego 7. Arlington, Virginia January 9, 2009 3 Hosted by Arlington County 8. Atlanta, Georgia January 22, 2009 5 Hosted by ARC 9. San Antonio, Texas March 4, 2009 3 Hosted by San Antonio-Bexar 10. Boise, Idaho March 18, 2009 3 Hosted by Kittelson Associates 11. Portland, Oregon March 19, 2009 4 Hosted by Kittelson Associates Task 3. Data Collection The objective of this task was to perform the data collection necessary to apply the 3-70 MMLOS method in each jurisdiction. The Philadelphia and San Diego area data collection efforts were completed on the same day as the workshop. The Florida test arterials data collection was funded by FDOT and completed by their contractor under their supervision. The data collection efforts for the remaining cities were conducted by NCHRP 3-70 team members and the host agencies prior to the day of the workshop. The data collection efforts are described in more detail in the attached summaries of the arterial testing by metropolitan area. Task 4. Analysis The objective of this task was to estimate the existing level of service for each selected street within each jurisdiction. Agency staff entered the data into the MMLOS spreadsheet under the supervision and guidance of members of the NCHRP 3-70 research team. The MMLOS spreadsheet then estimated the LOS using 4 different auto models (the recommended NCHRP 3- 70 stops model, the alternative NCHRP 3-70 speed model, the current HCM 2000 model, and the proposed NCHRP 3-79 percent free-flow speed model, plus the NCHRP 3-70 recommended transit, bike, and pedestrian models. For San Antonio, Texas two additional variations on the bike model and two additional variations on the pedestrian LOS model were tested. Details of these tests are provided in the attached test summaries.
Page 5 Exhibit 2: Field Test Results City Street HCM Auto LOS 3-70 Auto LOS 3-70 Transit LOS 3-70 Bike LOS 3-70 Ped LOS Arlington Wilson F D A F C Glebe D C B E D George Mason C B F D B Atlanta Bullsboro D B F E E 17th B B B B C Buford C B B F E Cobb C B C F F Boise Capitol E D C D D Broadway E B A E D State A B C E E Gainesville Archer A B E E D 13th St C B C E D University C B C D C Tower B B E E C Philadelphia Spruce D E F D B Broad C B A E B Chestnut E B A C A Portland Burnside D C C D D 39th Ave E C B F D 185th Ave F C B C C Powell B B A C C San Antonio San Pedro A B A F D Zarzamora B B C E D Broadway B B A E D Basse C F F E E San Diego Broadway E D A D B India A B F D C Tallahassee Capital Circ A B F E D Macomb A B F D C Tennessee B B A E C Appleyard A B D D C Tampa Himes C F D F D Nebraska D B B E E US 41 A B F E E Kennedy A B A D D Level of service âEâ and âFâ results are shaded.
Page 6 Task 5. Assessment The objective of this task was to obtain local agency assessments of the 3-70 MMLOS method (data requirements, analytical methods, and results). Detailed summaries of participant comments are provided in the workshop reports attached. The workshops suggested the strong need for improved guidance on the use of the methodologies. Users not accustomed to multimodal LOS analysis were put-off by the data collection requirements. Information on variable defaults, data collection short cuts, and sensitivities would be an extremely valuable addition to the model guidance. FDOT has already done much to simplify data needs for its agencies by developing software, reducing the range of conditions that can be evaluated and embedding defaults in the software. Florida and Portland believed that the bike LOS results should have been better than what they obtained using MMLOS. San Diego thought the MMLOS bike LOS results were too good, but that was for a case of bikes using a bus street where the buses stopped in the only available travel lane in each direction. Florida users, accustomed to the FDOT bike and pedestrian models, were puzzled why the overall facility LOS produced by the MMLOS models were so different at times from the individual segment and intersection results produced by the FDOT models upon which the MMLOS models were based. The driveway interference (for bikes) and roadway crossing difficulty (for pedestrians) are two new factors in the NCHRP 3-70 MMLOS bike and pedestrian models that were not present in the original FDOT bike and pedestrian segment and intersection models. These new factors can cause the overall LOS for the whole street to be worse than FDOT LOS for the individual segment and intersection components. Portland was concerned that their bike boulevards might produce an unrealistically low bicycle LOS in the MMLOS because of the numerous low volume, low speed driveways on these streets. This potential problem could conceivably be solved by combining a series of low volume, low speed driveways into a single high volume driveway for the purposes of coding the MMLOS model inputs. For example, 10 single family residential driveways in the field might be considered to be the equivalent of 2 standard driveways for the purposes of the MMLOS model inputs. Reactions to the four auto LOS models varied. The results of the 4 models were within 1 level of service of each other in the preponderance of field tests (see attached evaluation of the auto model results by Dr. Nagui Rouphail). Portland had one case where the auto speed LOS and the auto stops LOS diverged significantly. This happened on one street with closely spaced signals where there were few stops measured in the field, but speeds in the field were significantly below the free-flow speed. It is a short segment with long delays at the downstream signal. Thus, only one stop was measured, but the average speed was quite low.
Page 7 Agency personnel from Florida, Portland, and San Diego expressed a strong preference for keeping the current HCM 2000 Auto LOS model. San Diego was concerned about explaining the new auto LOS model to developers currently undergoing their development review process. Task 6. Refinement The objective of this task was to make the refinements to the MMLOS method, its Userâs Guide, and the software engine identified in the previous task. The majority of the refinement work was on the software engine implementation of the MMLOS method. However a few methodological refinements have also been made to improve the midblock pedestrian crossing delay computation. Two new versions of MMLOS (version 7 and 8) were made in November 2008 in response to comments received from the Florida users of MMLOS version 6, and in response to comments raised at the Philadelphia workshop. The following refinements to the methodology were implemented in version 8 of MMLOS.xls: 1. Reduced the arterial crossing distance used in the computation of the roadway crossing difficulty factor to exclude the shoulder/parking lanes, bike lanes, and median. 2. Changed guidance for applying roadway crossing difficulty factor. It is no longer recommended that it be turned off if jay-walking is illegal. It is now user discretion whether or not to include it in the Pedestrian LOS computation. 3. Corrected the computation of the traffic volume for pedestrian and bike LOS. This was a software error, not a methodological error. 4. Corrected computation of delay for pedestrians crossing midblock. Original method assumed that mean pedestrian wait time at a signal was equal to half the wait time per cycle. The correct value is one half of the wait time per cycle squared divided by the cycle length. 5. Defined the distance between intersections as being from stop bar to stop bar to be consistent with HCM 2010 definition. Delay at downstream signal to get to downstream cross bar is included in upstream segment. Questions from the NCHRP 3-92 contractor about the computations for the bicycle segment LOS were addressed and resolved. The NCHRP 3-92 contractor also identified some computational inconsistencies (described below) which were addressed in version 9 of MMLOS. The following refinements identified by the NCHRP 3-92 contractor and others identified by the NCHRP 3-70 team were incorporated in Version 9: 6. Added new pedestrian space computation (area per pedestrian). Needed for next fix below. 7. Fixed discrepancy between pedestrian density and non-density LOS numerical equivalencies. (identified by NCHRP 3-92 contractor) 8. Added two stage pedestrian midblock crossing delay computation if median is 6 feet or greater.(employs method suggested by NCHRP 3-92 contractor) 9. Corrected incorrect bus PTTR parameters for CBD and non-CBD (identified by NCHRP 3-92 contractor) 10. Fixed transit Fh equation to match users guide (identified by NCHRP 3-92 contractor)
Page 8 11. Macros written and added to automate printing, updating, and clearing (identified by NCHRP 3-70 team) 12. Removed references to external spreadsheets (identified by NCHRP 3-70 team) 13. Fixed error in computation of average stops per mile over whole arterial (identified by user) 14. Various cosmetic improvements to improve usability of software engine (identified by users and the NCHRP 3-70 team). The following refinements were added to version 10 of the MMLOS spreadsheet in late January 2009 to facilitate data entry and provide warnings about a few user coding errors: 15. Changed âfeet per treeâ fields to ânumber of treesâ 16. Added warning message if Buffer space is zero and trees are non-zero 17. Added warning message if parking lane is <=7 feet and parking occupancy is non-zero 18. Added warning message if Buffer is zero and trees are non-zero 19. Added warning message if parking lane is <=7 feet and parking occupancy is non-zero 20. Changes to several fields to trap for divide by zero problems if less than 5 segments present 21. Changed pedestrian multistage crossing formula to use values in "parameters" sheet. The formula now tests for minimum median width and type. The following refinements were added to version 10.3 of the MMLOS spreadsheet in late March 2009 in response to Boise and Portland workshop comments. 22. Fixed bad reference in segment 5 to left turn type for auto los computation. 23. Changed description of cross-street data entry field from 2-way vph to 1-way vph/lane to match definition of variable for pedestrian intersection LOS computation 24. Changed bike and pedestrian segment LOS reports to show segment and intersection results separately, The pedestrian and bicycle facility LOS is now shown only for total facility 25. Added check to pedestrian density computation for zero or negative ped flow rates 26. Highlighted all data entry fields with orange background (user request). 27. Added ability to select among the 4 auto LOS models for reporting auto LOS results A special version 11 of the MMLOS spreadsheet was created for Sprinkle Consulting on March 1, 2009 to test alternative bicycle and pedestrian LOS models. The results of those tests are reported in their attached summary of the San Antonio field tests. The following refinements were added to version 10.5 of the MMLOS spreadsheet in late April, early May 2009 to address issues identified by the NCHRP 3-92 contractor and Sprinkle Consulting regarding the bicycle and pedestrian LOS computations: 28. Removed duplicate application of peak hour factor to auto volumes in bike intersection and segment LOS. 29. Removed duplicate application of peak hour factor to auto volumes in ped segment LOS. 30. Fixed computation of average effective width for bike LOS to match eqn 31, page 83 Report 616. 31. Added Check for Less Than Zero effective width for bike LOS model.
