Active Transportation and Demand Management (ATDM) Analytical Methods for Urban Streets

CHAPTER 8. PROJECT CONCLUSIONS AND RECOMMENDATIONS

Building on recent studies and research inside and outside the Highway Capacity and Quality of Service Committee (HCQSC), this project provided new urban street active transportation and demand management (ATDM) methodologies, data sets, and content for the Highway Capacity Manual (HCM). This content may subsequently be incorporated within other chapters throughout the HCM. To the extent possible the measures and data used in this work were based upon existing sources, methodologies, and projects. In other words, major new data collection was not performed for this project.

The HCM is one of the most widely-used transportation documents, and the signalized analysis procedures are the most heavily utilized procedures in the Manual. The HCM complements traffic simulation tools; which require more time, expertise, and resources. By continuously developing new guidelines and updated procedures to address new facilities and technologies, the HCQSC hopes to support congestion mitigation in a time of population growth and limited resources. The recent release of HCM-based ATDM analysis procedures is an important element of the U.S. Department of Transportation's technology transfer mandate; in that it enables these new technologies and control strategies to be intelligently analyzed and adopted by more cities, and by the broader professional practice community. The HCM provides an ideal vehicle to disseminate these capabilities at a level of analysis that most engineers and planners are familiar and comfortable with. However new research is needed to develop HCM-compliant macroscopic relationships, targeted at urban street ATDM analysis.

Many urban street ATDM strategies were described in Chapter 1. Based on stakeholder (i.e., peer review group) feedback, the team's knowledge of HCM methods, the review of HCM-based freeway ATDM research, and the review of urban street ATDM strategies, the research scope was narrowed to three specific ATDM strategies: adaptive signal control, reversible center lanes, and dynamic lane grouping. The research team believes there would be significant national interest in developing the ability to analyze impacts of these strategies in an HCM context, and without requiring micro-simulation. These strategies could be classified as active transportation management (ATM) strategies as opposed to ATDM strategies, because they aim to move traffic more efficiently but without demand management.

It was desired at the outset of the project to develop robust models based on existing field data, while using simulated data to fill in the gaps. However due to the scarcity of field data located during Task 4 – Existing Data Sources, it became clear that any models developed during this project would need to be based on simulated data, with field data used for validity checks at best. Existing data were only found for the same three ATM strategies chosen to be the top priorities of this project. Moreover, only adaptive signal control had a sufficient quantity of existing data studies to support model development. As such, HCM-compatible models for reversible and/or dynamic lanes would be heavily dependent on simulation studies, to be conducted during Task 6 – Original Research. The original simulation studies from this project provided a detailed set of ranges and conditions under which the dynamic lane grouping strategy could be effective. It also provided an extensive set of potential benefits from dynamic lane grouping and adaptive signals.

In the early stages of the project, it was believed that the three chosen ATM strategies could be effectively modeled via capacity adjustment factors, similar to what was accomplished during the freeway ATDM project. However, it was later discovered that the capacity adjustment paradigm would be unsuitable for arterials, and that the HCM reliability framework would offer a preferable solution. Specifically, the alternative lane use configurations could be modeled as special event datasets within the HCM reliability framework, along with re-optimized timing plans for the new lane uses. The inadequacy of the capacity adjustment paradigm was detailed in Chapter 5.

Although capacity adjustment data collected during the dynamic lane grouping (DLG) experiments was not ultimately useful, delay reduction data collected during those same experiments produced the following observations:

1. Significant DLG benefits only occur when turn movement degree of saturation (volume to capacity, or v/c) exceeds 95 percent; and when adjacent through movement degree of saturation is at least 5 percent lower than (N-1)/N, where N is the number of exclusive through lanes prior to DLG treatment.
2. Benefits are much higher when g/C is high on the DLG approach and/or movement prior to DLG treatment (in other words, when DLG movements receive green during most of the cycle prior to DLG treatment).
3. Benefits are increased when turn movement degree of saturation is in the mid 100 percent range, prior to DLG treatment (in other words, when adding a new turn lane allows the turn movement to go from significantly oversaturated to undersaturated).
4. Benefits are increased when cycle lengths can be re-optimized to accommodate the new lane grouping (may not be effective with other congested intersections in the urban street, which may govern the background cycle length).
5. Similar benefits are observed for left turns versus right turns, and for two exclusive through lanes versus three exclusive through lanes.
6. Benefits are increased when there is good progression quality on the DLG approach, prior to DLG treatment.
7. Shared lane DLG produces no significant benefits for right turns unless RTOR is allowed, and produces no significant benefits for left turns.
8. If it were possible to automatically detect v/c and green time to cycle length (g/C) of various turn movements, it might become more effective to implement DLG in real time than by fixed time-of-day.

The adaptive signal simulation experiments produced the following observations:

• Together with the field study results obtained for Task 4, these simulations provided the type of data that could potentially be used to develop analytical models for the HCM.
• Although adaptive control produced consistently positive impacts according to a wide range of performance measures, the magnitudes of these benefits were less consistent.
• Adaptive signal benefits were often more significant under rainy conditions, a finding consistent with prior separate studies.
• Although adaptive control did not improve average performance under snow conditions, because speeds were generally very low for all vehicles, it did improve performance reliability under these conditions.
• Potential benefits in comparison to an optimized semi-actuated plan are still unknown.
• The impacts of high-priority input parameters (i.e., Table 3 from Chapter 4) are still unknown.
• Although it would be possible to develop case studies demonstrating the impacts observed during this project, there is still no clear path to developing a generalized adaptive signal model for the HCM, which would accurately handle a wide range of real-world conditions.

The reversible center lane (RCL) experiments produced the following observations:

• RCL benefits are generally maximized when degrees of saturation and directional splits are both maximized.
• The tipping point at which RCL benefits are viewed as significant may depend on a large number of additional factors affecting urban street operations (e.g., signal spacing, progression quality, distribution of major/minor-street demands, corridor reliability).
• HCM reliability analysis could be an effective method of predicting RCL benefits for a wide variety of urban street conditions.

Case studies for dynamic lane grouping and reversible center lanes provided a proof-of-concept for ATM strategy analysis within the HCM reliability framework. It is not clear whether adaptive signals or other advanced ATDM strategies could be accurately analyzed in this manner. DLG case study results also implied that in an oversaturated, multi-period model, not all time periods need to satisfy the pre-requisite criterion suggested in list item #1 on the previous page.

Follow-on studies should perform a more comprehensive set of ATM strategy experiments in the HCM reliability framework. HCM computational engines should have their special event modeling capabilities updated, so that ATM strategies can be modeled in a 24-hour manner when desired. Other ATM strategies based on relatively simple geometric concepts, similar to dynamic lane grouping and reversible center lanes, should be analyzed via the special event dataset method. A more extensive research effort might be needed for bringing adaptive signals into the HCM framework, although the secrecy of adaptive signal algorithms may continue to be an obstacle. Additional research would also be needed for bringing demand management strategies (e.g., congestion pricing) into the HCM framework; although this might require additional progress in the surface-freeway HCM integration effort, or in using HCM methods to model complex grid networks.