Incorporating Travel Time Reliability into the Highway Capacity Manual (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.

111 C h a p t e r 8 This chapter recommends the next steps to carry forward the results of this research. The following topics are addressed: • Defining reliability levels of service; • Implementing the L08 research; • Identifying freeway facility research needs; and • Identifying urban streets research needs. Defining reliability Levels of Service Reliability can be used to define level of service (LOS) in a variety of ways. The intent of this section is to identify a pre- ferred approach for defining LOS based on the concept of reliability. LOS definitions require that cutoff points (bound- aries) of the measurement unit be established that define each LOS range. The research team established the ranges described in this section to be illustrative, not definitive. The final selection of ranges will require additional analysis and, more important, a vote and approval by the Highway Capac- ity and Quality of Service Committee. The research team also proposed one option for defining reliability LOS (with input from the project panel), but the decision to adopt a formal definition lies with the Highway Capacity and Quality of Service Committee. All reliability measures should be consistently scoped for accurate comparisons between facilities. Reliability measures created from observations during peak periods or peak hours vary greatly from measures created using 24-hour observa- tions. The temporal extent of the analysis relative to the peak period(s) therefore critically affects the values of reliability measures. Facility selection (i.e., the definition of start and end points of the facility relative to the critical bottleneck) also plays a part in the reliability measures. Options for Reliability LOS Options for defining reliability-based LOS fall into four categories: • Freeway reliability based on current LOS ranges; • Freeway and urban streets reliability LOS based on travel speed ranges; • Freeway reliability LOS based on most-restrictive condi- tion; and • Freeway and urban streets reliability LOS based on the value of travel. Each option is discussed in more detail in the following subsections and illustrated using example data from various U.S. facilities. Option 1: Freeway Reliability LOS Based on Current LOS Ranges The simplest method for defining reliability LOS is to use the existing LOS definitions for freeway facilities and basic freeway segments—based on density—and urban streets and urban street segments—based on travel speeds. For each facility type, the analysis procedure produces a distribution of the LOS measure that represents the per- centage of trips (or the percentage of analysis periods) that fall into each LOS range. Alternately, the definition can be based solely on the percentage of trips (or percentage of analysis periods) in LOS F alone. Table 8.1 shows examples of this approach using freeway detector data from Seattle, Washington, and Atlanta, Georgia. Both the peak (north- bound direction, in both cases) and off-peak directions are shown. For example, if reliability is based solely on the percentage of trips operating under LOS F, then the Recommendations

112 southbound approach for the Seattle facility is “unreliable” 11.1% of the time (i.e., its reliability is 88.9%). Similar dis- tributions can be constructed for urban streets and signal- ized intersections using the current LOS ranges in the HCM2010. Density-based LOS for freeways is a significant departure from the concept of travel time reliability. The primary issue is that travel times do not vary much over a wide range of density-based LOS ranges. Further, by using the current free- way LOS ranges, the focus is on unsaturated (uncongested) conditions: LOS A through E is in the unsaturated range while all oversaturated conditions are grouped into a single LOS F category. Finally, the lower-density thresholds for weaving sec- tions further complicate the use of density as the fundamental measure of reliability. For urban streets, the definition would be consistent with the measurement of travel times (travel speeds over a distance of highway are derived from travel times). Travel times are not relevant for signalized intersections since they are “points” on the highway system; but for highway segments, delay is related to travel times as well. Option 2: Freeway and Urban Streets Reliability LOS Based on Travel Speed Ranges In this approach, travel speed ranges can be constructed for freeways in a manner similar to that for urban streets. Here, travel speed is analogous to space mean speed (SMS) over the entire freeway facility or segment. The LOS ranges may be based on percentages of the free-flow speed as they are for urban streets, or they may be set at fixed SMS values. Table 8.2 shows the values used for urban streets. Because of the insensitivity of travel speeds to a wide range of density and volume-to-capacity (v/c) values (current LOS A through D), one option is to extend the number of LOS ranges for oversaturated conditions. An example of how this method would be applied is shown in Table 8.3, again using detector data from Seattle and Atlanta. Under this option, the south- bound direction for the Seattle facility would be at LOS F (or “unreliable”) 17.8% of the time. Table 8.1. Density-Based Reliability LOS Based on Free-Flow Speed LOS Facility Density (pcpmpl) Percentage of Trips in Each LOS Range, Weekdays, 4:30–6:00 p.m. Seattle, I-405 (2007) Atlanta, I-75, Northside (2010) NB SB NB SB A ≤11 0.0 0.0 0.9 1.1 B >11–18 8.1 1.4 0.4 1.9 C >18–26 91.6 32.9 1.6 5.8 D >26–35 0.3 34.0 2.7 69.8 E >35–45 0.0 20.6 4.7 18.4 F >45 or d/c > 1.0 0.0 11.1 90.7 3.0 Mean TTI 1.016 1.352 1.984 1.050 Note: pcpmpl = passenger cars per mile per lane; NB = northbound; SB = southbound; d/c = demand-to-capacity ratio. Table 8.2. LOS Ranges for Urban Street Facilities and Segments Travel Speed as a Percentage of Free-Flow Speed (mph) LOS by Critical v/c Ratio <–1.0 >1.0 >85 A F >67–85 B F >50–67 C F >40–50 D F >30–40 E F ≤30 F F Note: v/c = volume to capacity. Source: HCM2010, Table 16-4 (TRB 2010a). Table 8.3. Freeway Reliability LOS Defined by Travel Speed or Travel Time Index Ranges LOS Travel Speed (mph)a and Equivalent TTI Percentage of Trips in Each LOS Range, Weekdays, 4:30–6:00 p.m. Seattle, I-405 Atlanta, I-75, Northside NB SB NB SB A ≥60 (TTI ≤ 1.083) 27.7 1.6 0.4 15.4 B 50–59 (1.083 < TTI ≤ 1.300) 71.9 48.3 6.6 80.5 C 45–49 (1.300 < TTI ≤ 1.444) 0.3 12.0 3.2 1.4 D 40–44 (1.444 < TTI ≤ 1.625) 0.0 9.3 8.5 0.3 E 35–39 (1.625 < TTI ≤ 1.857) 0.1 11.0 14.4 0.8 F <35 (TTI > 1.857) 0.0 17.8 66.8 1.6 Mean TTI 1.016 1.352 1.984 1.050 Note: NB = northbound; SB = southbound. a Average speed over the length of the facility (i.e., the space mean speed).

113 Option 3: Freeway Reliability LOS Based on Most-Restrictive Condition Options 1 and 2 are predicated on providing a distribution of the percentage of time the facility is unreliable, rather than assigning a single “grade” to define a highway’s LOS. Although the distribution is highly useful for analysts, it may confuse nontechnical audiences who are used to using a single LOS value. Focusing on the percentage of trips/analysis periods in LOS F, rather than specifying the percentage in each LOS range, is a departure from how LOS ranges are defined in the HCM2010. An alternative is to report the most restrictive condition. In this approach, travel speed boundaries are again defined, but the observation value is the percentage of trips greater than or equal to each travel speed. An example is shown in Table 8.4. This approach requires setting a second threshold value: the cumulative percentage of trips for the restrictive condi- tion. The analyst reads down the table, starting from LOS A and finds the LOS at which the cell value is greater than or equal to the threshold. For example, if the analyst wants 75% of trips to be in the most restrictive range, the data in Table 8.4 yields the following results: • Seattle I-405 NB: LOS B; • Seattle I-405 SB: LOS E; • Atlanta I-75 NB: LOS F; and • Atlanta I-75 SB: LOS B. This option is functionally equivalent to selecting a per- centile value for a threshold and seeing where it falls in the table. In Table 8.4, using the value at which 75% of trips fall is equivalent to calculating the 75th percentile TTI, convert- ing it to travel speed (i.e., space mean speed), and finding the range in which it falls. Alternately, instead of using the 75th percentile TTI, the 25th percentile SMS can be used. Table 8.5 shows the data for this method. A variant on this approach is to use a reliability metric for the threshold values. This option also establishes reliability LOS on the basis of a single value but in a simpler manner. It establishes LOS ranges for a reliability metric and makes the assignment solely on the basis of where the facility’s calcu- lated value falls. For illustration, the following example uses the planning time index (PTI) as the reliability metric; this is the ratio of the 95th percentile travel time to the free-flow or “ideal” travel time. Given the ranges shown in Table 8.6, the reliability LOS for the Seattle and Atlanta analysis sections can be assigned. For example, the LOS for the southbound direction of the Seattle facility is F. Table 8.4. Freeway Reliability LOS Based on Most-Restrictive Condition LOS Travel Speed (mph) Cumulative Percentage of Trips in Each Range, Weekdays, 4:30–6:00 p.m. Seattle, I-405 Atlanta, I-75, Northside NB SB NB SB A ≥60 27.8 1.6 0.4 15.4 B ≥50 99.6 49.9 7.0 95.9 C ≥45 99.9 61.8 10.3 97.3 D ≥40 100.0 71.2 18.8 97.7 E ≥35 100.0 82.2 33.2 98.4 F ≥30 100.0 92.2 51.2 99.5 Mean TTI 1.016 1.352 1.984 1.050 Table 8.5. Percentile Values Matched to Speed Ranges Section 75th Percentile TTI Corresponding Travel Speeda (mph) LOS (from Table 8.4) Seattle, I-405 NB 1.018 58.9 B Seattle, I-405 SB 1.545 38.8 E Atlanta, I-75 NB 2.338 25.7 F Atlanta, I-75 SB 1.037 57.9 B Note: NB = northbound; SB = southbound. a Based on free-flow speed of 60 mph. Table 8.6. Freeway Reliability LOS Based on the Planning Time Index LOS PTI Calculated PTI, Weekdays, 4:30–6:00 p.m. Seattle, I-405 Atlanta, I-75, Northside NB SB NB SB A 1.00–1.10 1.032 B >1.10–1.25 1.155 C >1.25–1.50 D >1.50–1.75 E >1.75–2.00 F >2.00 2.129 3.035 Mean TTI 1.016 1.352 1.984 1.050 Note: NB = northbound; SB = southbound.

114 Generally speaking, all SMS measures can be converted to TTI for consistency. LOS can be defined on the basis of the full distribution of TTI; on the fraction of time TTI exceeds a given value (associated with LOS F); or on the basis of a range at a specified TTI percentile—for example, the 75th, 80th, or 95th percentile. Option 4: Freeway and Urban Streets Reliability LOS Based on the Value of Travel A somewhat radical departure from traditional LOS philoso- phy, which is based on performance from the perspective of the facility, is to base LOS on the value of travel perceived by users. The concept is to translate the values of both typical (average) travel time and travel time reliability into travel time equivalent values and then assign a cost to them. The LOS ranges are then based on unit costs per traveler. This approach can be applied to both interrupted and uninterrupted facilities. The valuation approach is based on the work of Small et al. (2005). They define the reliability ratio as the value of reli- ability (VOR) divided by the value of time (VOT). SHRP 2 Project C04 suggests a range of 0.5 to 1.5, but a review of past studies suggests that it is more in the 0.9 to 1.2 range. There- fore, a value of 1.0 seems reasonable for composite trips. However, previous research also indicates that the value of reliability varies by trip purpose. Because the research to date has been limited, the value of the reliability ratio is still uncer- tain. Small et al. (2005) adopted the quantitative measure of variability as the upper tail of the distribution of travel times, specifically, the difference between the 80th and 50th percen- tile travel times. They argue that this measure is better than a symmetric standard deviation since, in most situations, being late is more crucial than being early, and many regular travel- ers build a safety margin into their departure time to leave them an acceptably small chance of arriving late (i.e., plan- ning for the 80th percentile travel time means arriving late only 20% of the time). On this basis, travel time equivalents can be defined and used to put both typical (average) and reliability components into the same units. That is, reliability is equilibrated to average travel time. The calculation of travel time equivalents is shown in Equation 8.1. ( )= + ×TT TT TT – TT (8.1)80 50ae m where TTe = travel time equivalent on the segment or facility; a = reliability ratio (VOR/VOT), set equal to 1.0 for now; TTm = mean travel time; TT50 = 50th percentile travel time; and TT80 = 80th percentile travel time. Table 8.7 shows the results of applying this procedure. The end result is an estimate of an equivalent delay value, normal- ized to segment length (delay per mile). The LOS ranges can then be set on delay per mile. This approach has the advantage of creating a single com- posite value for facility performance. In addition to deviating from traditional HCM LOS philosophy, the nascent nature of reliability valuation research is a problem in that future work is likely to produce different calculation methods and reli- ability ratios. Summary of Options Option 1: Reliability LOS Based on Current LOS Ranges This option is the most consistent with current LOS concepts in the HCM2010. For urban streets and urban street seg- ments, the current LOS ranges based on travel speeds can be used to present an LOS distribution (percentage of trips in each LOS range). For freeways, basing reliability LOS on the current density- based LOS designations is not useful. Travel times are most variable under congested conditions, and the current density ranges for LOS A through E do not result in much change in travel times. Table 8.7. Travel Time Equivalents and Equivalent Delays for Use in Setting Reliability LOS Section Travel Time (min) Delay per MileMean 80th Percentile 50th Percentile TTe Free-Flow Travel Time Excess Travel Time (delay) I-405, NB 5.95 5.98 5.94 5.99 5.77 0.22 0.038 I-405, SB 6.67 8.13 5.91 8.89 4.94 3.95 0.800 I-75, NB 10.90 13.55 10.53 13.92 5.51 8.41 1.526 I-75, SB 6.13 6.07 5.97 6.23 5.84 0.39 0.067 Note: NB = northbound; SB = southbound.

115 However, creating a distribution rather than a single LOS value can be difficult to communicate to nontechnical audi- ences (a major use of the LOS concept). A simple solution is to report only the percentage of trips in LOS F or E + F, but this misses the remainder of the LOS distribution. While a single value is clearly more consistent with HCM2010 prac- tices, the use of a distribution appears to lend itself to a reli- ability analysis. A reliability analysis inherently captures a range of operating conditions on the same facility and attri- butes those conditions to various sources of (un)reliability. Using a distribution of LOS values therefore intrinsically mir- rors the variability of traffic conditions on the facility. Option 2: Freeway Reliability LOS Based on Travel Speed Ranges This option would make freeway reliability LOS conceptually consistent with urban streets and urban street segments. The problem of presenting a distribution rather than a single LOS value is still present. Option 3: Freeway Reliability LOS Based on Most-Restrictive Condition This method avoids the problem of presenting a distribution and assigns a single LOS value. It is more complicated to apply and explain in that two values must be set: a percentage threshold for the trips that fail to meet LOS criteria and the ranges for each LOS category. Option 4: Reliability LOS Based on the Value of Travel This option is the most complicated to both develop and explain. It has the advantage of being based on travelers’ per- ception of reliability, but it relies on a factor (the reliability ratio) that has not been precisely identified and will likely change with new research. In addition to its complexity, establishing LOS ranges on the basis of travel time equiva- lents is highly problematic. Recommended Option Testing the four options with field data did not reveal a clearly better choice on which to base reliability LOS. Further, the research team found the four options to be difficult to com- municate to the profession, the public, and decision makers. As a result, the team decided to develop an “on-time” measure similar to Option 2. This measure, the reliability rating, is the percentage of trips served at or below a threshold TTI (the ratio of actual travel time to free-flow travel time). The selected thresholds are 1.33 for freeways and 2.50 for urban streets. These thresholds approximate the points at which most travel- ers would consider a facility congested; thus, the measure roughly reflects the percentage of trips on a facility that experi- ence conditions better than LOS F. The difference in threshold TTI values results from differences in how free-flow speed is defined for freeways compared with urban streets, as TTI is measured relative to free-flow speed. The research team has not defined a service measure for travel time reliability. Because travel time reliability is a new concept for the transportation profession, the research team recommends that performance measures be used to describe the travel time reliability performance on freeways and urban streets. Subsequently, consideration can be given to using travel time reliability to define LOS. When reliability is con- sidered as a service measure, the team recommends that the reliability rating (now a performance measure) be the basis. Other considerations for future reliability LOS delibera- tions follow. Urban Streets Figure 16-4 of HCM2010 defines LOS F as either (1) where travel speed is 30% or less of the base free-flow speed or (2) where the subject through movement at one or more inter- sections has a v/c ratio greater than 1.0. Because the LOS defi- nition is based on travel speed, which is a derivative of travel time, no changes in the LOS concept for urban streets is needed. Freeways For freeway reliability, the research team first recommends that the existing density-based LOS definition be replaced with a travel speed–based definition. Density should be maintained as the indicator of general freeway performance, especially for rural facilities. The team also recommends that, at some point in the future, travel speed be considered as a replacement for density even for general performance on urban facilities. The use of travel speed as the indicator of both general and reli- ability performance on freeways also provides consistency with the urban streets method. Implementing the L08 research The draft HCM reliability chapters and computational engines (FREEVAL-RL and STREETVAL) were completed in draft form in the fall of 2012. The materials were fully vetted and reviewed by the TRB Highway Capacity and Quality of Service Commit- tee in conjunction with the 2013 TRB annual meeting. The computational engines consist of spreadsheets with embedded Visual Basic code. Separate Excel spreadsheet tools are used to generate the scenarios and run the FREEVAL and

116 STREETVAL engines, thus executing the HCM2010 calcula- tions in an automated fashion and processing the results for reliability reporting purposes. Although not part of the L08 project, a natural extension of the computational engines and other tools would be the development of a more user-friendly, integrated software tool that could execute the files faster than the Excel-based computational engines. Such a software tool could be hosted on a fast server and located in any secure environment, including a cloud-based environment. Cur- rently, the updated FREEVAL and new STREETVAL compu- tational engines are hosted in the developer’s environment at the contractor’s site. No decision has been made on the host- ing arrangement for the final product. Identifying Freeway Facility research Needs Research needs in the freeway facilities methodology take into consideration improvements to the core HCM2010 methodology and to the reliability submodels developed in the course of this study. Research to Overcome Core Methodology Limitations While the freeway facility methodology has been significantly improved and expanded in the course of this study, it still needs additional research to fill some significant gaps. Oversaturated Model The oversaturated flow-density relationship has not been cali- brated since its inception in the 2000 HCM. Several research efforts have compared the results of the HCM2010 method with field observations in an effort to validate the predicted performance. Nonetheless, a more rigorous calibration effort is desired, with the potential of enhancing the current linear speed–flow relationship used to model operations at densi- ties greater than 45 pcphpl. Off-Ramp Spillback Modeling Spillback from off-ramps is not considered in the current methodology, significantly weakening its ability to model congested corridors. Off-ramps are often choke points along freeway facilities, especially in the case of a freeway–arterial corridor pair, as a signalized intersection may cause spillback onto the freeway. The current implementation uses only a simple off-ramp capacity check, without further scrutiny of the impacts on the freeway. An enhanced off-ramp spillback model should be sensitive to queuing patterns on the freeway, which are likely to use only some of the freeway lanes (depending on the cross-section), and should also consider speed drops in the adjacent nonqueued lanes. Free-Flow Speed and Capacity Effects The free-flow speed and capacity adjustment factors used throughout the methodology to account for nonrecurring congestion effects have been adopted from the most recent and relevant literature, but they have not been locally cali- brated or validated. While a literature synthesis is an appro- priate approach for a project like this, it carries the risk of inconsistencies in parameter definitions and data collection. A coordinated research effort would allow for a consistent evaluation of these effects and should carry a special empha- sis on interaction effects, such as inclement weather in work zones, or incidents on inclement weather days. Managed Lane Modeling While the methodology developed in this project does not explicitly incorporate the new method for analyzing man- aged lanes completed under NCHRP Project 3-96, analysts may use the 3-96 results to calibrate this methodology’s base facility inputs. Specifically, the 3-96 method introduces the concept of two parallel lane groups on a freeway facility, dis- tinguishing between general purpose and managed lanes. The 3-96 method generally does not change the underlying methodologies for general purpose lanes; rather, it empha- sizes the development of speed–flow curves and friction effects for managed lanes. The methodology is modeled so that managed lane operations affect general purpose lanes in cases of “cross-weave friction” resulting from at-grade access points to and from the managed lanes. Analysts wishing to perform a reliability analysis on the general purpose portion of a managed lane facility should calibrate the base facility performance by using the 3-96 method and, to implement additional capacity adjustment factors in the L08 method, should the presence of access points result in significant friction impacts on the general purpose lanes. The analyst would run the base facility seed file in both the FREEVAL-RL and the FREEVAL-ML engines, and then calibrate the performance of FREEVAL-RL to match the FREEVAL-ML friction effects. Research to Improve the Reliability Submodels Demand Impacts of Nonrecurring Congestion Research is needed to understand and quantify the effects of weather, work zones, and special events on traffic demand. The current demand variability is a function of the day of the week

117 and the month of the year; as such, it accounts for some implicit correlation between, for example, weather and demand (e.g., more snow in the winter and generally lower traffic demands). However, no explicit modeling has accounted for the impacts of weather, work zones, and special events, although they are intuitively expected to reduce demand through diverted trips, carpooling, and other effects. Similarly, work zones are expected to affect facility demand. Depending on the level of penetration of traveler information systems, even incidents may result in demand shifts. Intuitively, all of these effects have an impact on the reliability of the facility and should be con- sidered in future research. Ignoring those sensitivities may result in overestimating the impact of non recurring conges- tion on reliability performance measures. Conditional Probabilities of Submodels The L08 method assumes that incident rates and weather conditions are independent. The method does account for the possibility of incidents during inclement weather events but assumes a simple multiplication of the underlying prob- abilities. Research is needed to develop models that explain the relationship and to derive conditional probabilities of incidents under different weather conditions, as well as inci- dents in work zones. Enhanced Weather Detail The methodology does not currently account for weather events that have a small effect on segment capacity reduction (<2%). In addition, a given weather event (e.g., rain, snow) is always assumed to occur at its mean duration value; and only two possible start times for weather events are considered. Although the low-capacity impact scenarios are incorporated in the general “nonsevere” weather category, the methodol- ogy would benefit from added detail on weather duration and weather starting times—both of which can be explored in future research. Enhanced Incident Detail To consider the average effect of incidents on a facility, the analyst assumes each incident is located on one of three pos- sible segments: the first segment, the segment at the facility midpoint, or the last segment. Similarly, the timing of each incident is set as either the start of a study period or its mid- point. Finally, only three possible incident durations are con- sidered: the 25th, 50th, and 75th percentiles of the incident duration distribution. These assumptions on incident loca- tion, starting time, and duration were essential in enabling the team to enumerate a discrete (and manageable) number of reliability scenarios. Future research may explore options for stochastic incident modeling in a reliability context, using segment-specific incident probabilities. Identifying Urban Streets research Needs The research conducted for this project has led to the formu- lation of several recommendations for future research on urban streets methodology. That research is grouped into two categories. The first category describes the research needed to overcome known limitations in the scope of the urban streets reliability methodology. The second category describes research needed to improve specific models within the reli- ability methodology. Research to Overcome Limitations In general, the urban streets reliability methodology can be used to evaluate the performance of most urban street facili- ties. However, the methodology does not address some events or conditions that occur on some streets and influence their operation. These events and conditions are identified in the following paragraphs. Calculation of Facilitywide Performance Measures The HCM2010 urban streets methodology predicts the travel time and speed of through vehicles (i.e., vehicles traveling along the facility and served as a through movement at each intersection). However, it does not describe a procedure for aggregating the performance of all movements on each seg- ment and that of all movements on each external intersection approach. This type of estimate would describe facilitywide performance and include measures such as vehicle miles trav- eled (VMT), vehicle hours traveled (VHT), and vehicle hours delay (VHD), and their equivalents for person movement (PMT, PHT, and PHD). Facilitywide performance measures are critical for invest- ment and planning studies that compare alternatives and select the most cost-effective alternative. These measures are also critical building blocks for a future HCM corridor methodol- ogy. The use of PMT, PHT, and PHD would facilitate assess- ments of the service provided to transit passengers, bicycle riders, and pedestrians, as well as auto drivers and passengers. The extended HCM2010 urban streets method should address speed estimates for nonthrough vehicles on the seg- ment (which are not currently covered in the HCM2010). In addition, the extended method should quantify the impacts of vehicles denied entry to the facility during each analysis period because of severe congestion within the facility. These effects could be quantified in terms of total VHD and PHD during the entire study period.

118 Truck Pick-Up and Delivery Lane and shoulder blockages resulting from truck pick-up and delivery activities in downtown urban areas can be con- sidered like incidents in terms of the randomness of their occurrence and duration. The dwell time for these activities can range from 10 min to 20 min. They are estimated to result in 950,000 VHD on the nation’s urban arterial streets (Chin et al. 2004). Research is needed to quantify the effect of truck pick-up and delivery activities on the speed and capacity of an urban street segment. The research should also develop models for predicting the frequency and duration of such activities. The scope of the research may be broadened to include the effect of on-street parking on urban street operation. The research results should be suitable for reliability evaluation. Signal Malfunction A signal malfunction occurs when one or more elements of the signal system are not operating in the intended manner. These elements include vehicle detectors, signal heads, and controller hardware. A failure of one or more of these ele- ments typically results in poor facility operation. For example, a detector failure typically causes a fail-safe operation in which a continuous call is held by the detection system, thereby extending the subject phase to its maximum green limit. A failure to a signal head or the controller hardware can result in a flashing-red operation, making traffic control equivalent to all-way-stop control. A failure to the communications system can lead to loss of signal coordination. Anecdotal information indicates that between 10% and 20% of an agency’s detection sensors are not functioning at any particular time. Research is needed to quantify the effect of signal malfunc- tion on the operation of an urban street segment. The research should focus on the more common types of malfunction. It should separately quantify the effect of each type on speed, saturation flow rate, and other traffic characteristics that influence urban street operation. The research should also develop models for predicting the frequency and duration of common types of malfunction. The research results should be suitable for reliability evaluation. Railroad Crossing and Preemption Chin et al. (2004) used data from the Federal Railroad Admin- istration to estimate the nationwide delay related to railroad crossings. Their evaluation considered all crossings on urban principal arterials, regardless of whether they occurred at a sig- nalized intersection. They estimated the delay to be 2,700,000 vehicle hours. This amount is relatively small in the context of other sources of urban street congestion (e.g., incidents, weather, work zones). Nevertheless, train crossing times can be lengthy (typically 5 min to 10 min) and can result in consider- able delay at each crossing. A railroad crossing at a mid-segment location on an urban street facility effectively blocks traffic flow while the train is present. Urban street operation can also be disrupted when a train crosses a cross-street leg of a signalized intersection. Sig- nal coordination may be disrupted for several cycles follow- ing train clearance. Research is needed to quantify the effect of railroad cross- ings on urban street operation. The research should address both mid-segment crossings and intersection preemption resulting from a cross-street crossing. The research should develop a procedure for quantifying the effect of train events on speed, saturation flow rate, and other traffic characteristics that influence urban street operation. The research should also develop models for predicting the frequency and dura- tion of train crossings and preemption events. The research results should be suitable for reliability evaluation. Adverse Weather Conditions The current methodology does not address weather conditions that restrict driver visibility or degrade vehicle stability. These conditions include fog, dust storms, smoke, and high winds. Chapter 10 of the HCM2010 indicates that a significant visibil- ity restriction can reduce freeway capacity by about 10%. It also indicates that high winds can reduce freeway capacity by 1% to 2%. The impacts of these conditions on urban street operation are unknown but are likely to be similar. Chin et al. (2004) esti- mated that fog results in 3,400,000 VHD on the nation’s urban arterial streets. Research is needed to quantify the effect of fog, dust storms, smoke, and high winds on urban street operation. The research should develop a procedure for quantifying the effect of these weather conditions on speed, saturation flow rate, and other traffic characteristics that influence urban street operation. The research should also develop models for predicting the fre- quency and duration of the associated weather events. The research results should be suitable for reliability evaluation. Research to Improve Specific Models The urban streets reliability methodology was developed using currently available data and research publications. The data were used to calibrate the various models that make up the methodology. Calibration data were also collected in the field when existing data were not available. In some instances, the research team noted that a model’s reliability could be improved if additional data were collected or made available through sub- sequent research. The following paragraphs identify research that targets these specific models in the methodology.

119 Wet-Pavement Duration The findings from one research project indicated that the time required for pavement to dry following a rain event is a function of temperature. Drying time was found to decrease with increasing temperature. However, the research did not consider drying time for temperatures below about 60°F. Research is needed to develop a model for predicting pave- ment drying time for temperatures ranging from -10°F to 100°F. Other factors that influence drying time (e.g., relative humidity, time of day, cloud presence, wind speed, and pave- ment type) may be considered in developing the model. Effect of Weather on Signalized Intersection Saturation Flow Rate A limited amount of research has investigated the effect of weather on saturation flow rate. This research found that weather events (i.e., rain or snow) reduce saturation flow rate, but the amount of reduction appears to vary depending on other, unmeasured factors. For example, the amount of reduc- tion may be influenced by driver experience in (or familiarity with) driving in poor weather. Research is needed to develop a saturation flow rate adjustment factor for the following weather conditions: • Clear, dry pavement; • Rain, wet pavement; • Clear, wet pavement (not raining); • Snow, snow or ice on pavement; and • Clear, snow or ice on pavement (not snowing). The research should quantify the effect of precipitation rate, grade, and temperature on saturation flow rate. Driver familiarity with the listed weather conditions should also be considered, possibly incorporating this effect using surro- gates such as the altitude and latitude of the intersection. Effect of Incident Length on Segment Operation As described in Chapter 6, several proposed enhancements were developed for urban streets methodology in Chapter 17 of the HCM2010. One is a procedure for quantifying the effect of a mid-segment incident on segment speed and capacity. This procedure does not consider the length of roadway influenced by the incident. An incident that closes a lane for the length of the segment likely has a larger negative effect on operation than one that closes the same lane but for only a few feet along the street. When an incident’s influence length is “short,” relative to the length of the segment, its proximity to the upstream or downstream signalized inter- section may also have an effect on segment operation. Research is needed to quantify the effect of incident length, duration, and location (relative to the adjacent signals) on segment speed and capacity. Incident Distribution Research indicates that the distribution of incident frequency varies in a predictable manner by the following categories: street location (i.e., segment or intersection), event type (crash or noncrash), lane location, and severity (i.e., fatal, injury, property damage only, breakdown, debris). However, research also indicates that the distribution proportions vary by region of the country (possibly explained by weather, terrain, income level, and design standards). They may also vary by the facili- ty’s degree of recurring congestion and its geometric design (e.g., presence of roadside barrier). Research is needed to develop a model for predicting the distribution of incident frequency that is suitable for nationwide application.