Improved Analysis of Two-Lane Highway Capacity and Operational Performance (2018)

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.

NCHRP 17-65 Improved Analysis of Two-Lane Highway Capacity and Operational Performance Final Report 1 1. Introduction 1.1. Background Two-lane highways account for a very significant portion of the national highway system and serve an essential function for the movement of people and goods. Measured in centerline miles, two- lane highways constitute the vast majority of the highway system in the United States. On these highway facilities, a single lane is provided for travel in each direction, resulting in a higher level of interaction between vehicles traveling in the same direction, and often the opposing direction as well. Specifically, maintaining a desired speed is dependent on the ability to pass slower vehicles, which in turn is a function of oncoming traffic level and available sight distance. The interaction between vehicles is also expected to increase with the increase in traffic flow and speed variation associated with a heterogeneous traffic mix. These operational characteristics typically result in formation of platoons and make the platooning phenomenon an important indicator of performance on two-lane highways. Most two-lane highways exist in rural areas and are generally characterized by low traffic volumes, relatively high speeds and lower design standards compared to those used on well- travelled multilane highways. However, as urban areas continue to see growth further away from the central cities, two-lane highways in previously less developed areas are experiencing increases in traffic demand. Additionally, as urban area congestion continues to build, shipping companies are more frequently considering less congested two-lane highways in their routing decisions. The presence of commercial trucks on two-lane highways poses additional challenges for maintaining acceptable levels of operational performance due to more variance in the geometric design of these facilities and less favorable passing opportunities. Although adding additional lanes to a two-lane highway will often address operational deficiencies with two-lane highways, such construction projects are very expensive. Having good and accurate analysis methods for two-lane highways may allow roadway design and traffic engineers to identify ways to make significant improvements to the operational performance of a two-lane highway without resorting to a full multilane configuration. The standard reference in the U.S. for traffic analysis techniques is the Highway Capacity Manual (HCM). The HCM, 6th edition, contains a chapter that provides an analysis methodology for two-lane highways. Unfortunately, the HCM analysis procedure falls short in several respects of providing roadway design and traffic engineers the methods they need for performing accurate and comprehensive two-lane highway facility evaluations. The current HCM analysis procedure has been criticized on several issues, which are described in the following section. 1.2. Computational Deficiencies in, and Key Gaps in Coverage of, HCM Analysis Methodology This section describes the computational deficiencies in, and key gaps in coverage of, the current (2016) HCM analysis methodology for two-lane highways.

NCHRP 17-65 Improved Analysis of Two-Lane Highway Capacity and Operational Performance Final Report 2 1.2.1. Speed-Flow Relationship The current relationship between speed and flow given by the HCM is linear in form, as shown in Figure 1-1. Figure 1-1. HCM 2010 speed-flow relationship Source: HCM 2010, Exhibit 15-2 This is inconsistent with the relationship identified in several studies (Luttinen, 2000; Brilon and Weiser 2006; Hammontree, 2010), where it has been found that the shape of the speed-flow relationship follows a concave up form, such as that illustrated in Figure 1-2. Figure 1-2. Speed-flow relationship for German two-lane highways (under ideal roadway design conditions)* * “VF” refers to average travel speed of passenger cars; “q” refers to flow rate (veh/h); “SV” refers to heavy vehicles Source: German HBS 2015 (German equivalent of HCM)

NCHRP 17-65 Improved Analysis of Two-Lane Highway Capacity and Operational Performance Final Report 3 Furthermore, even if a linear or approximately linear speed-flow relationship were to be used, it is more logical that the slope of this linear relationship should be affected by free-flow speed (FFS) and the directional distribution of traffic, as illustrated in Figure 1-3. As for multilane facilities, average speed at capacity tends to converge to a narrow range of values, assuming higher speed facilities in which the posted speed limit does not constrain free-flow speeds. The availability of passing opportunities in the oncoming lane and the relative proportions of directional flow rates also affect the rate of change in average speed. (a) traffic in one direction only (b) 50/50 directional traffic split Figure 1-3. HCM speed-flow curves (dashed lines) versus speed-flow curves with slope as a function of FFS (solid lines) Source: Luttinen et al., 2005 1.2.2. Service Measures Several researchers such as Luttinen (2000, 2001, 2005), Catabagan and Nakamura (2009), Al- Kaisy and Freedman (2011), Van As (2003), Polus and Cohen (2009), and Morrall and Werner (1990) have suggested that the HCM two-lane highway analysis methodology is very simplistic and largely inaccurate for level of service (LOS) assessment and that Percent Time Spent Following (PTSF) and Average Travel Speed (ATS) are not necessarily the most appropriate

NCHRP 17-65 Improved Analysis of Two-Lane Highway Capacity and Operational Performance Final Report 4 performance measures to use for assessing two-lane highway performance. Both PTSF and ATS are used to evaluate Class I two-lane highways (high-speed, major inter-city routes and connectors of major traffic generators). PTSF is used solely to evaluate Class II two-lane highways (lower speed, intra-city routes). In the HCM 2010, a third class of two-lane highways (low speed, in moderately developed areas) was added, for which the service measure is Percent Free Flow Speed (PFFS). Ideally performance measures should (Luttinen et al., 2005): 1. Reflect the perception of road users on the quality of traffic flow. 2. Be easy to measure, estimate, and interpret. 3. Correlate to traffic and roadway conditions in a meaningful way. 4. Be compatible with the performance measures of other facilities. 5. Describe both uncongested and congested conditions. 6. Be useful in analyses concerning traffic safety, transport economics, and environmental impacts. It can be argued that PTSF may satisfy conditions 1 and 3. However, it is difficult to measure in the field, it is not compatible with the service measures of other facilities, it does not describe the extent of congestion, and it is not very useful in other analyses. PTSF is also not a good performance measure for indicating if improvements should be made to a highway that has low volumes with a high percentage of heavy vehicles and few passing opportunities. In this case, PTSF would be high and suggest that additional lanes or passing sections need to be added even though the volumes are low. The second measure, ATS, is only used on class I two-lane highways (may not be relevant on Class II and class III highways). Unlike PTSF, ATS is easy to measure in the field, if point measurements are sufficient, otherwise vehicle matching or floating vehicle surveys are required. Although ATS is easy to measure in the field, it is not very informative about the efficiency of the highway. Since the analysis section of a two-lane highway facility is usually several miles long, there could be many changing conditions, such as posted speed limit and roadway alignment that affect ATS, yet it is not related to varying traffic conditions. This makes ATS somewhat meaningless for determining how the highway is operating (Al-Kaisy and Freedman, 2011). PFFS is meant to account for the limitations of ATS. It measures the speed reduction due to increased traffic volume and/or platooning, which makes it possible to compare the current conditions to the ideal conditions (Al-Kaisy and Freedman, 2011). Volumes in the analysis direction and opposing direction were found to be statistically significant in the PFFS models developed by Al-Kaisy and Freedman (2010). One of the limitations of PFFS is that it is almost unaffected by the addition of a passing lane, which indicates that it is not a useful performance measure for capturing the delay caused by platooning (Al-Kaisy and Freedman, 2010). The HCM 2010, however, currently only uses PFFS for determining the LOS of Class III highways, which do not include passing lanes, by definition. The service measures should also be conducive to integrating two-lane highway segments with signalized and unsignalized intersections to support a facility-based analysis (i.e., long stretch

NCHRP 17-65 Improved Analysis of Two-Lane Highway Capacity and Operational Performance Final Report 5 of a two-lane highway with a number of segments with varying characteristics and occasional intersections). For a corridor or area-wide analysis, it is also desirable to be able to integrate the two-lane highway analysis with the analysis of freeways and multi-lane highways. 1.2.3. Deterministic Method for Identifying Vehicle Follower Status For purposes of measuring PTSF in the field, the HCM indicates that a value of 3 seconds or less can be used as a surrogate for identifying vehicles in a following mode. That is, if the headway between successive vehicles is 3 seconds or less at a chosen point along the highway facility, the trailing vehicle is considered to be in a following state, for the purpose of estimating PTSF. This method does not account for drivers having different desired following headways. However, a few empirical studies (Al-Kaisy & Durbin, 2008; Dixon et al., 2002; Luttinen, 2001) found that the HCM models in estimating PTSF do not reasonably relate to the aforementioned method for measuring percent followers in the field using the 3-second headway rule. Specifically, the HCM PTSF estimates were found to be consistently and significantly higher than the corresponding field-measured percent followers. Catabagan and Nakamura (2009) found that the 3-second headway criterion used for the PTSF measure in the HCM underestimates the number of following vehicles. They found their proposed probability-based follower identification method to more accurately describe traffic conditions on Japanese highways. 1.2.4. Truck Passenger Car Equivalent (PCE) Values The PCE values developed for the two-lane highway analysis methodology in the HCM were based on results from the TWOPAS simulation program (Harwood et al., 1999). The PCE values for two-lane highways differ from the values for other highway types and, even within the two- lane highway procedure, different sets of PCE values are needed for different performance measures: average travel speed and percent time spent following. The PCE values vary as a function of terrain and directional flow rate. In the HCM 2000 (TRB, 2000) version of the analysis methodology, the truck PCE values were a function of flow rates in units of passenger cars. This created an awkward iterative approach to determining the PCE values since the PCE values are needed to convert a measured flow rate in units of vehicles to one in units of passenger cars. There was also a situation where one could oscillate between PCE values and be caught in an “endless loop”. In the HCM 2010, the units for the PCE look up tables were changed from passenger cars to vehicles to avoid this awkward situation, but this change was not based on any new research or theory. Additionally, it is necessary to use the measured hourly volume divided by the peak hour factor with the grade adjustment factor and PCE value tables, which is somewhat awkward. The PCE values in the HCM are average values that do not vary with the percentage of trucks in the traffic stream, even though it has been shown (Luttinen, 2001a) that the first few trucks added to the traffic stream have a much greater operational effect than subsequent increments of truck percentage. That is, the PCE is a function of the heavy vehicle (HV) percentage itself. It is also possible that for some conditions a PCE value does not even exist; that is, the mixed flow of passenger cars and trucks may have a lower ATS than any (uncongested) passenger car flow (Luttinen, 2001a).

NCHRP 17-65 Improved Analysis of Two-Lane Highway Capacity and Operational Performance Final Report 6 This issue is described further in the subsequent paragraphs and figure (1-4). Because of these difficulties, the German capacity manual does not use PCEs (Brilon & Weiser, 2006). Recent research on the effects of trucks on freeway operations under the NCFRP 41 project (Dowling et al., 2014) suggests that the standard HCM PCE approach of converting a mixed flow rate of passengers cars and heavy vehicles into an equivalent passenger car only flow rate cannot produce the observed speeds of trucks and passenger cars on extended upgrades (grades in excess of 2% extending 1/2 mile or longer). There is no point on the passenger car speed flow curve that can produce the observed crawl speeds of trucks on extended upgrades, and when volumes or the percent trucks become high, the ability of passenger cars to pass trucks decreases until all vehicles are moving at the speed of the slowest truck climbing the grade. Extrapolating these results to a two-lane highway upgrade with no passing, it suggests that the speeds of all vehicles will rapidly reduce to the crawl speed of the slowest truck in the traffic stream rather than to an equivalent passenger car speed at a higher flow rate. Figure 1-4 shows a simulated freeway speed-flow pattern from NCFRP Report 31 (Dowling et al., 2014) for a 5-mile long 6% upgrade and compares that to the idealized passenger car speed- flow curve for a 70 mi/h freeway in Chapter 11 of the 2010 HCM. As can be seen there is no single capacity-based PCE value that can map the HCM passenger car curve to the observed average speed pattern for a 30% truck mix. A capacity-based PCE will get the analyst to point “A” on the graphic, but the predicted speed from the passenger car curve is 25 mi/h too high. A second speed-based PCE would be needed to arrive at point “B”, the actual mixed flow speed at capacity. There are several additional problems with the PCE approach that are illustrated in Figure 1-4. The spread of average mixed flow speeds is much greater at lower flow rates, which is the opposite of what happens in an idealized passenger car only traffic stream. The average speeds of traffic at low flow rates are unstable on a long upgrade because of the interference with slow moving trucks. In addition, a PCE that works for capacity will not be appropriate for lower mixed flow rates.

NCHRP 17-65 Improved Analysis of Two-Lane Highway Capacity and Operational Performance Final Report 7 Figure 1-4. Divergence of passenger car and mixed flow speed-flow patterns on freeway upgrade Adapted from Exhibit 59, NCFRP Report 31 (Dowling et al., 2014) and Chapter 11, 2010 Highway Capacity Manual. 1.2.5. Poor Guidance on Estimating Base Free-Flow Speed (BFFS) According to the HCM, no specific guidance can be given on the estimation of the "base" free- flow speed due to the broad range of speed conditions. It does state that “a very rough estimate of BFFS might be taken as the posted speed limit plus 10 mi/h”. (TRB 2000, 2010) The posted speed is used as a rough starting point because FFS is based mainly on the roadway geometry. Grade severely affects traffic flow on highways and the HCM does not address those effects very well (Luttinen, et al., 2005). The adjustment factors for grade only give a rough estimation of the geometric conditions. Horizontal curves can also have a significant impact on FFS. ATS calculations are significantly affected by the BFFS. 1.2.6. Overestimation of Performance Measure Improvements Due to Passing Lanes The results obtained from the inclusion of a passing lane lead to unrealistic improvements to the performance measures and, consequently, the level of service. PTSF and ATS are very sensitive to changes in volume, and passing lane adjustments result in large improvements to both PTSF and ATS. When solving for a service volume, which is the volume corresponding to a given LOS, based on PTSF and ATS for a facility with a passing lane, the calculated volume is unreasonably higher than the volume for the facility with no passing lane (Washburn, 2011). These results are misleading because, realistically, the number of vehicles before and after a passing lane section should be approximately the same. A small percentage of the vehicles may change positions, but it does not drastically change the number of vehicles in the traffic stream.

NCHRP 17-65 Improved Analysis of Two-Lane Highway Capacity and Operational Performance Final Report 8 This means a passing lane should not show major increases to the traffic-carrying capability of the highway section. Furthermore, the HCM two-lane analysis methodology chapter provides conflicting guidance on climbing lanes—it indicates that adding a passing lane to a segment that is operating at LOS F will not improve LOS; thus, there is no need to perform a passing lane analysis. However, an upgrade segment can be the bottleneck of the highway. If a passing lane is added on an upgrade segment (i.e., climbing lane), which is followed by a level segment, it is conceivable that the LOS could be improved. 1.2.7. Capacity The HCM 2010 states that directional capacity is 1700 pc/h and two-way capacity cannot exceed 3200 pc/h (TRB, 2010). Capacity is computed by adjusting the base flow rate of 1700 pc/h for grade and heavy vehicles. Implicit in these values is that the capacity of the observed direction is reduced when the opposing flow rate exceeds 1500 pc/h. Some empirical studies have examined two-lane highway capacity (e.g., Rozic 1992; Luttinen, 2000; Brilon and Weiser, 2006; Kim, 2006), but the literature is generally sparse with such studies. Theoretical studies (references provided on p. 145 of Luttinen, 2001a) have indicated that the influence of the opposing flow does not change significantly when the opposing flow rate has increased above 400–450 veh/h. Earlier studies (references provided on p. 145 of Luttinen, 2001a) do not give support to the assumption that the opposing flow has an effect on capacity only after the opposing flow rate has exceeded 1500 pc/h. Of the capacity values reported, as well as the factors affecting these values, there is considerable variance. Finnish studies have indicated that there is a capacity drop under congested conditions. As traffic density exceeds optimum density (approximately 20–25 veh/km) the maximum flow rate is approximately 300–400 pc/h lower than capacity (see Figure 1-5). This information is valuable in those rare cases when a two-lane highway is congested; for example, during weekend peaks, work zone arrangements or incidents. One major challenge, however, to determining capacity values with much accuracy is that it is difficult to find two-lane highway sites that operate at or near capacity. Transportation agencies typically convert two-lane highways to multilane highways once they reach intolerable levels of PTSF and platooning, which usually occur at flow rates well below capacity.

NCHRP 17-65 Improved Analysis of Two-Lane Highway Capacity and Operational Performance Final Report 9 Figure 1-5. Speed-flow values on Finnish arterial highway 4 (grade separated intersections and passing lanes) to the north of Lahti Source: Luttinen et al., 2005 1.2.8. Facility Scope The current analysis methodology is limited to the segment-level. Ideally, the methodology would be such that it can be integrated with signalized and unsignalized intersection analyses, as well as account for multiple contiguous two-lane highway segments with differing attributes (passing zone, passing lane, etc.) to allow for a facility-/corridor-level analysis. As mentioned previously, one of the challenges to expanding the analysis of two-lane highways to include intersections and other highway segment types is the compatibility of service measures. Yu and Washburn (2009) and Li and Washburn (2014) addressed this issue by using a delay-based measure across two-lane highway segments and signalized intersections. Compatibility of service measures between two- lane highways and other uninterrupted-flow facilities (multilane highways, freeways) would also be desirable. 1.2.9. Ease of Use There are several aspects to the current HCM two-lane highway analysis methodology that are awkward for users. As previously mentioned, PCE values that vary by service measure, and especially the iterative approach that was present in the HCM 2000 (which was removed in the HCM 2010 in an attempt to reduce user confusion, but without documentation of its effect on the accuracy of the method), often cause confusion for users. Although this is less of an issue when

NCHRP 17-65 Improved Analysis of Two-Lane Highway Capacity and Operational Performance Final Report 10 using a software implementation of the methodology, experience by research team members in teaching this methodology has consistently shown this issue to be confusing to students. The current HCM analysis methodology chapter is heavy on the use of adjustment factor tables. While a considerable number of adjustment factors is reasonably expected, these adjustment factors are implemented exclusively in tabular format. And for each of these tables, it is usually necessary to perform an interpolation, and in a couple of cases, a three-way interpolation is often necessary. The methodology should move away from the tabular implementation of adjustment factor values to the extent possible and directly implement equations. This will increase the user-friendliness of the methodology and also make software implementation easier. 1.3. Simulation Tools In addition to the need for a two-lane highway analysis procedure for the HCM that improves upon its various deficiencies and limitations, the transportation engineering and roadway design profession could benefit from a modern microscopic simulation tool that includes the ability to model two-lane highways, particularly the phenomenon of passing in an oncoming lane. Given the complexity of some two-lane highway facilities, it is unreasonable to expect that a deterministic, analytic procedure will be capable of accurately analyzing all two-lane highway configurations. For complex situations that are not amenable to analysis with the new HCM procedure, analysts should be able to utilize a simulation tool to help them analyze these situations, such as is commonly done for arterial and freeway corridors. Unfortunately, simulation tools that were commonly used for modeling two-lane highway facilities in the past are no longer viable tools for future applications, for myriad reasons that are described later. 1.4. Research Objectives and Scope The objective of this research was to (1) identify appropriate performance measures for operational and capacity analyses of two-lane highways and develop models to produce these performance measures in an HCM context, and (2) develop or modify a microsimulation-based analysis method for two-lane highways. The resulting methods for the analysis of two-lane highways are suitable for incorporation into a future edition of the HCM.