Incorporating Truck Analysis 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.

28 S e c t i o n 4 The Transportation Research Information Database (TRID) (TRB) was scanned to identify publications within the past 10 years that are relevant to the key topics of this project. The key topics investigated in this literature review were • Treatment of effects of other modes on trucks in the HCM 2010, in other guides internation- ally, and in research; • Treatment of truck effects on other modes (i.e., automobile, bus, bicycle, pedestrian) in the HCM 2010, in other guides internationally, and in research; • Truck classification schemes—weight-based, axle-based, and length-based; • Integration of trucking needs into transportation investment decisionmaking—typical plan- ning studies involving freight movement (case studies); • Key highway performance criteria critical to shippers and carriers; and • Sources of data on truck movements (which will be important to future users of the HCM truck analysis method). The literature review also investigated how the effects of trucks on other modes and the effects of highway performance (and other modes) on trucks are treated in major highway capacity manuals worldwide. In addition to the United States’ Highway Capacity Manual (TRB, 2010), the team evaluated highway capacity guides from several countries, including • Canada’s Capacity Guide for Signalized Intersections (Canadian Institute of Transportation Engineers, 2008); • The United Kingdom’s Design Manual for Roads and Bridges (Department of Transport, Highways Agency); • Australia’s Guide to Traffic Management (Austroads); • Germany’s Highway Capacity Manual (FGSV, 2001); • India’s Road Congress Guidelines (Indian Roads Congress, 1994); and • The Indonesian Highway Capacity Manual (Indonesian Directorate General of Highways, 1993). 4.1 Critique of the HCM 2010 This section evaluates the 2010 Highway Capacity Manual (HCM) (TRB, 2010) from two perspectives: its ability to predict the specific performance of trucks and its ability to model the effects of trucks on the traffic stream. Regarding the ability of the HCM to predict the specific performance of trucks: • The HCM does not provide methods for estimating truck speed and performance as distin- guished from that of the passenger vehicle stream. Literature Review

Literature Review 29 • The HCM methods, MOEs selected, LOS thresholds, and so forth are based on expert opin- ion and not validated with freight drivers except for the Urban Streets Method (Chapter 17). The method does not identify measures of effectiveness that reflect the perspectives of motor vehicle shippers and carriers. • The HCM does not currently predict travel time reliability—a performance measure of criti- cal concern to shippers and carriers. • There is no specific truck LOS methodology in the 2010 HCM. With the exception of the Urban Streets Methodology, where adjustments are made to the saturation flow rate, trucks are treated using an adjustment factor to the demand (by means of passenger car equivalent [PCE] factors). Regarding the ability of the HCM to model the effects of trucks on the traffic stream: • The HCM truck classification scheme is extremely simplistic, not reflecting the spectrum of truck performance capabilities in the U.S. fleet. • The HCM PCEs are too simplistic since they do not reflect the variation in the truck fleet or the influence of truck proportion or grades on urban street PCEs. • The HCM PCE look-up tables stop at 25% trucks (as a percentage of total traffic flow) even though there are many facilities in the United States where trucks routinely exceed 25% and can exceed 50% of the average daily traffic flow. • The HCM treatment of trucks is inconsistent across chapters. Most notably, the uninterrupted flow chapters use a volume-to-PCE conversion method early-on in the procedure. Many inter- rupted flow chapters use PCEs directly to adjust methodology parameters (e.g., saturation flow rate adjustment for signals). 4.1.1 How the HCM Currently Models Trucks The 2010 HCM defines three heavy-vehicle types: transit buses, recreational vehicles (RVs), and trucks. These three types are grouped in the HCM under the broader category of heavy vehi- cles. A heavy vehicle is defined in the HCM as “A vehicle with more than four wheels touching the pavement during normal operation.” In the HCM, buses, recreational vehicles, and trucks are considered heavy vehicles with the following special characteristics: • A bus is defined as “A self-propelled, rubber-tired road vehicle designed to carry a substan- tial number of passengers (at least 16) and commonly operated on streets and highways.” (Equivalent to FHWA Class 4.) • A RV is defined as “A heavy vehicle, generally operated by a private motorist, for trans- porting recreational equipment or facilities. Examples include campers, motor homes, and vehicles towing boat trailers.” (Generally equivalent to FHWA Class 5.) • A truck is defined as “A heavy vehicle engaged primarily in the transport of goods and materials or in the delivery of services other than public transportation.” (Equivalent to FHWA Classes 5–13.) Each heavy-vehicle type can be assigned its own PCE rating for the purposes of capacity and operational analyses. However, in most cases, the HCM groups these vehicle types together. Buses and trucks are usually assigned identical PCE values while RVs are generally assigned a slightly lower PCE value. The value of PCEs ranges from an assumed fixed number (e.g., signalized inter- sections, roundabouts) to a detailed PCE-estimation procedure (e.g., uninterrupted flow chapters). In general, most HCM chapters convert heavy vehicles to equivalent passenger car units and add them to the passenger car volumes to obtain the total equivalent passenger car volume that is used in the HCM methodologies. The HCM analysis then estimates the capacity, density, speed, delay, and LOS for the equivalent passenger car stream. Truck speeds and delays are not isolated from the values predicted using the equivalent passenger car stream performance. From

30 incorporating truck Analysis into the Highway capacity Manual a performance perspective, none of these covers weight-to-horsepower ratios, which are a key measure of performance. The impacts of trucks on other modes are modeled differently in the HCM according to the facility type: • On freeways, multilane highways, and two-lane highways, the grade affects the PCEs for trucks. The number of trucks affects the equivalent passenger car volume, which in turn affects the density and speed of traffic. • On urban streets, the PCE equivalent of trucks is independent of grade. The number of trucks affects the estimated saturation flow rate, which in turn affects speed and therefore automo- bile and transit LOS. • The HCM methodologies for estimating bicycle LOS are sensitive to the percentage of trucks in the traffic stream. For multilane and two-lane highways, the grade or general terrain affects the result by affecting the truck PCEs used to compute the average speed of the passenger car equivalent traffic stream. For urban streets, the percentage of trucks directly affects the bicycle LOS. • The HCM methodologies for estimating pedestrian LOS are indirectly sensitive to trucks. Higher truck volumes result in lower estimated average automobile speeds, which in turn positively affect pedestrian LOS. (This is no doubt an unintended side effect of excluding the direct effects of trucks in the pedestrian LOS model.) • The HCM methodologies do not explicitly account for the acceleration, top speed, headways, and climbing ability of trucks when assessing their impact on other modes. 4.1.2 Critique of How HCM Models Trucks This section critiques the weaknesses of the current HCM methodology for modeling the effects of trucks on the traffic stream and facility performance. Several areas of concern have been identified, including underestimation of truck traffic in freight-dominated corridors; sim- plistic PCE conversions, which homogenize all trucks into a single truck type; and a lack of estimates of truck value of time and travel time reliability. HCM Has Too Narrow of a Range for Truck Percentages The HCM 2010 does not have a methodology to incorporate truck percentages higher than 25% on certain freeway segments that have a long grade (greater than 1.0 or 1.5 miles)—for example, the HCM can only handle truck percentages of 25% for positive grades (uphill) and 20% on negative grades (downhill). While 25% is a significant amount of trucks, there are many facilities that have truck percentages higher than this (see Exhibit 8). I-81 in Virginia, for exam- ple, has truck demands of between 40% and 50% (Rakha et al., 2007). This is not an isolated situation, with 6% of all counting stations in California (Caltrans, n.d.) and 3% of all counting stations in Virginia (Virginia DOT, 2008) recording truck volumes greater than 25%. HCM PCE Conversions Too Simplistic The HCM 2010 interrupted flow chapters lump trucks and buses together in computing the PCE values of heavy vehicles. The HCM approach is also independent of significant variables like the truck type and weight-to-horsepower ratio. Rakha et al. (2007) found that while PCEs are constant for low grades (under 2%), they decrease with increasing truck proportions. They suggested that different PCE values would be appropriate as truck proportions increased. This concept of PCE values needing to be related to proportion was further supported by Webster and Elefteriadou (Webster, 1999), who developed recommended PCE values for truck percentages of up to 60% using a weight-to-power ratio of 137.5 lb/hp. Their recommended PCEs were further adjusted to account for other weight-to-power ratios.

Literature Review 31 Control delay at signalized intersections in the current HCM methodology does not consider the probability distribution of heavy vehicles within the queue; it only takes into account satura- tion flow rate using PCE factors. No provision is available to adjust start up loss or end gain with heavy vehicles in different queue positions. Ramsay et al. (2004) suggest this should be done in computing expected control delay. They propose incorporating a weighted average control delay considering vehicles in different queue positions. The uninterrupted flow chapters currently assume a reduction in PCEs for facilities with a high percentage of trucks. This is presumably because the traffic stream becomes more homogeneous, which reduces the marginal impact of each heavy vehicle. Although not quantified in the HCM, state controls—such as the implementation of electronic toll collection and weigh-in-motion technology—surely have an impact on the performance of heavy vehicles in the traffic stream. Elefteriadou et al. (2007) found that PCEs are highly sensitive to the weight-to-power ratio of heavy vehicles. Unfortunately, the HCM assumes a single truck type that was calibrated using a mix of trucks and buses with an average weight-to-horsepower (hp) ratio of between 125 and 150 lb/hp (see Chapter 11, HCM 2010). This does not account for the wide range of heavy vehicles on the road today—for example, both Middleton (2006) and Rakha et al. (2001) found a variety of trucks that have weight-to-power ratios greater than 150 lb/hp with values reaching Exhibit 8. Percent trucks on National Highway System. Source: FHWA Office of Freight Management and Operations, 2010.

32 incorporating truck Analysis into the Highway capacity Manual as high as 300 lb/hp. It is also important to consider that effective horsepower can vary consider- ably with various weather, pavement, and vehicle conditions. Primary Issues in Measuring Facility Performance Effects on Trucks There are a number of issues with using the Highway Capacity Manual to estimate how facility performance will affect trucks as perceived by motor vehicle carriers and shippers. These issues include that • Service measures used in the HCM are not relevant to carriers/shippers, • No provision is made in the HCM for evaluating truck LOS separately, and • The current HCM does not deal with travel time reliability for trucks. Service Measures in the HCM Are Not Relevant to Carriers/Shippers Service measures are a subset of performance measures, which are used to estimate LOS (letter grades A through F). With the exception of two-lane highways, the uninterrupted flow facility service measure is density. However, density is not a direct concern to shippers—even though density correlates to speed and delay (which is a concern for shippers), density itself is not a tangible service measure for movers of freight. Service measures for two-lane highways vary by facility type: Class I facilities2 use speed and percent time following, with Class II facilities relying solely on percent time following. Class III is the final type with its service measure based on the percentage of free-flow speed. Of the current measures, speed is the only service measure that may be particularly relevant to freight movers. The percentage of time that an automobile has to follow a truck is not a significant consideration for commercial goods movers. Interrupted flow facilities such as urban streets with signalized intersections use service mea- sures that do concern freight movers such as average control delay, but are not consistently sen- sitive enough to other measures like the number of stops on a route. Furthermore, these service measures are often quantified into LOS letter grades ranging from “A” to “F” that are based on the expert opinion of the authors of the chapters of the HCM (except Chapter 17: Urban Streets). Unfortunately, these perceptions to date have not reflected the perceptions of truck drivers, but rather are based on automobile drivers. No Provision in the HCM for Evaluating Truck LOS Separately The 2010 HCM does not have any methodology or LOS criteria specifically targeted to the needs of truck freight movers. Current HCM analysis methodologies convert trucks to equiv- alent passenger cars and then analyze the traffic stream as if it were all passenger cars. This approach results in a single traffic stream with averaged speeds, densities, and delays that does not fully consider the wide range of both vehicles and drivers on the roads today. While knowing average performance measures is a useful tool in traditional operational analy- ses, it would still be useful to have a way of measuring truck LOS as its own measure. This is espe- cially true given the economic importance of freight movement by truck. Unfortunately, there is no provision for evaluating truck LOS separately from automobile LOS in the HCM 2010. This lack of methodology becomes a critical gap when an agency is considering measures to restrict truck activity from certain facilities, lanes, or peak hours. The conventional automobile LOS analysis shows significant benefits for automobile drivers by using these measures, but there is no corresponding analysis of the disadvantages to trucks or the economic activity of the region. 2 Two-lane highway “classes” are meant to address differences in purpose and driver expectations: Class I highways are con- nectors, where high speed is valued; Class II highways are recreational or scenic routes in which high speeds are not expected nor desired; Class III highways serve moderately developed areas (they may be portions of a Class I or Class II highway).

Literature Review 33 HCM 2010 Does Not Deal with Travel Time Reliability A factor of great concern to movers of commercial goods is travel time reliability, which is especially important today because many businesses adhere to what is known as “just-in-time delivery,” so they do not have to have large warehousing facilities. This delivery method requires coordination and the ability to accurately determine shipping time. The 2010 HCM does not provide any methodologies for determining travel time reliability, nor are there any measures provided that can be relevant to reliability. There is major research being performed under the Strategic Highway Research Program (SHRP) in travel time reliability: SHRP2-L08 will add content to the HCM that deals with reli- ability and reliability performance measures. Incorporation of this research into the NCFRP Project 41 will be useful in developing truck LOS methodologies given that travel time reliability is such an important factor in freight movement. 4.1.3 Deficiencies of Current HCM Methodologies Based on the results of the literature review, a review of international practice, the agencies survey, and the shippers/carriers survey, the deficiencies of the current HCM 2010 methodolo- gies for evaluating the impacts of trucks on other highway users and the impacts of other high- way users on trucks are as follows: • The 2010 HCM PCE tables for specific grades on uninterrupted flow facilities stop at 25% trucks (e.g., HCM: Chapter 11). They should be extended to higher proportions of trucks in the traffic stream, perhaps to 50%. • The 2010 HCM PCE tables for uninterrupted flow facilities and fixed single PCE values for interrupted flow facilities have undocumented built-in assumptions as to the distribution of weight to equivalent horsepower ratios (WEHPR) and lengths of trucks in the traffic stream. This limitation requires the original research to be repeated every few years as the vehicle fleet evolves. A methodology to predict PCE as a function of the distribution of WEHPRs and length of trucks in the traffic stream would facilitate maintenance and updating of future HCM PCE tables as the vehicle fleet changes. • The current HCM heavy-vehicle categories of trucks, buses, and RVs needs to be subdivided for trucks into WEHPR and length subtypes to better match PCEs with actual vehicle mix and to produce more accurate estimates of the capacity effects of trucks. • Urban Street Method does not have a factor for truck effects on base free-flow speed or run- ning speeds between signals (HCM: Chapter 17). • There are no truck-specific performance measures producible with the current HCM meth- odology. The HCM produces only average traffic stream delays and speeds without isolating the delays and speeds for trucks separately. • The HCM does not identify a LOS measure specific to trucks. 4.1.4 Appropriate Uses of HCM for Truck Operations Analysis Many public agencies have typically used HCM methods to compute volume/capacity ratios, average speed, average delay, and automobile LOS for the mixed automobile and truck traffic stream. The agencies then assume that average results for automobiles and trucks can then be used as a proxy for the conditions experienced by trucks in the traffic stream. Where greater detail is needed, some agencies have used microsimulation analysis models to better model the interactions of trucks with each other and the passenger car traffic stream and report average traffic stream performance. Relatively little has been done to tease out truck-specific performance measures from the output of conventional HCM and microsimulation models.

34 incorporating truck Analysis into the Highway capacity Manual In economic analyses used to prioritize projects for investments, separate values of time (VOT) have been assigned to truck movements within the traffic stream. However, due to the lack of truck-specific performance measures available from conventional modeling methods, truck-specific VOTs have been applied to mixed traffic stream speeds and delays. The accuracy of the benefit/cost analyses currently used to prioritize transportation system investments can be significantly improved if the correct truck-specific performance measures were available: average truck speed, average truck delay, and reliability. The VOT for truck movements used to compute the economic value to the region of trans- portation investments would also be greatly improved if it could be specified in terms of a distri- bution rather than a single average value. A framework that could provide truck average speed, delay, and reliability according to the specific VOT for each subtype of truck movement would greatly improve the accuracy of the economic analyses used to prioritize transportation system investments. Truck LOS letter grades (i.e., A through F) keyed to truck delays, truck speeds, reliability, and the VOT for the specific truck movements being evaluated would also greatly facilitate the iden- tification of freight movement problem spots in the highway network, comparison of alternative improvements, the determination of significant impacts, and the determination of acceptable and unacceptable performance. 4.1.5 Recent Research on Truck Level of Service While separate truck LOS methodologies have not been incorporated into the HCM to date, there have been a number of research efforts that have been trying to deal with this issue. While these research efforts are an excellent foundation, much more is still needed before a working truck LOS methodology can be incorporated into the HCM. One issue is that these prior research efforts have surveyed primarily truck drivers rather than carriers or shippers. One can see the bias in the results caused by focusing on drivers rather than carriers and shippers. The truck driver responses during these interviews are more concerned about comfort and convenience of the trip rather than trip time or reliability. Hostovsky and Hall Hostovsky and Hall (2003) conducted a focus group study of truck drivers at the annual convention of the Ontario Trucking Association (OTA) in Canada. Members of the OTA Road Knights Team were used for the focus group. The OTA Road Knights Team is a group of 10 professional transport drivers with first-class driving records who make presentations on how car and truck drivers can safely share the road. The Road Knights Team consists of only tractor-trailer drivers and does not include drivers of straight trucks, dump trucks, buses, and other heavy vehicles. The goods carried by the OTA participants include office products (from Toronto to New York City), general freight around Toronto and to the United States from Ontario, and chemical tankers. Two simple, open-ended questions were used to start the focus group: 1. When driving your truck, what makes for a good trip on a freeway for you? 2. When driving your truck, what makes for a bad trip on a freeway for you? The participant responses were grouped under freeway conditions, traffic conditions, atti- tudes of other drivers, safety, and aggressive driver behavior. Travel time (or speed), traffic density (or maneuverability), and traffic flow were three major variables that were all mentioned with regard to freeway conditions quality of service.

Literature Review 35 For freeway conditions, drivers were worried about factors such as road marking in construction zones, narrower lanes, snow being cleared promptly, and evenness of pavements. Drivers tend to prefer the middle lanes, which offer the “smoothest ride.” For traffic conditions, a key issue men- tioned repeatedly was steady traffic flow. Due to the longer time required for trucks to accelerate and decelerate (compared with automobiles) drivers have a very negative perception of stop-and- go traffic. Drivers mentioned that “Traffic moving steadily within an acceptable range” was the most important factor to them rather than speed. This does not necessarily imply drivers do not value speed, but may rather reflect an implicit understanding from experience (of the drivers) that steady speed is more likely to result in a higher overall speed than stop-and-go traffic. Drivers point out that congestion due to construction, road maintenance, and accidents is very problematic for their operations—for example, while closures of major freeways at night for maintenance and construction do not bother commuters, it does affect truckers because it is considered “premium truck traffic time” where they can travel without interference from automobiles. The authors concluded that “truckers are concerned about travel times (or aver- age speed) and about maneuverability, but there was a stronger consensus on the importance of what they termed flow or moving at a steady speed within an acceptable range.” Another issue mentioned is attitudes toward other road users. Here truck drivers’ general perception was that they were more professional and consistent in their driving behavior and habits than were other road users. Their perception was that the driving behavior of other non- truck drivers (lane changing, signaling, etc.) was inconsistent and disconcerting. This behavior negatively impacted their perception of a trip. Safety was a recurring theme throughout the focus study that included driving behavior of other road users, maintenance of safe driving conditions (during winter), clearance of snow, rubbernecking at accident scenes, and so forth. The key finding of the study was that the very nature of the tractor-trailers (large, heavy vehicles with stiff suspensions that require long braking distances and more time to acceler- ate) makes them place significance on LOS variables other than traffic density, which is used by engineers and planners in the HCM. It is not traffic density that mattered to them—it was traffic “flow,” which means what mattered most to this particular group of participants was a “comfortable operating range of highway speeds that does not require much braking and gear changes related to acceleration. Most of the truckers do not mind reduced freeways speeds as long as the traffic is flowing steadily.” The primary concern of urban freeway commuters was travel time, especially reduced time in light of frequent stops. Therefore, it can be concluded that truckers do think of time for their ability to deliver cargo, but they prefer a longer predictable time to having an unsteady traffic flow when perceiving their quality of service. Washburn A preliminary methodology for assessing truck LOS only for basic freeway segments was developed by Washburn in 2002. This methodology was developed without benefit of surveys of motor vehicle carriers, shippers, or even truck drivers. The methodology proceeds upon the presumption that maneuverability is an important factor for truck drivers (Washburn, 2002). The methodology is based on a function of the ratio of percent of free-flow speed (FFS) of trucks to percent of FFS of passenger cars and is referred to as the “Relative Maneuverability Index,” or RMI. It can be expressed as Equation 2RMI Percent FFS Percent FFS trucks pcars = The intent of the RMI was to capture the effect in which truck drivers are not able to change lanes at the same frequency as cars do at various density levels. Thus, trucks experience lower

36 incorporating truck Analysis into the Highway capacity Manual average speeds than cars because they cannot perform as many discretionary lane changes to maintain their desired speed. This research did not perform measurements of this presumed preference of drivers to change lanes to maintain desired speed nor did this research attempt to measure or capture the perceptions of motor vehicle carriers or shippers. Conceptually, RMI approaches a value of one under both free flow and ideal geometric condi- tions when most cars and trucks are travelling at their desired (albeit different) FFS. It would also approach 1.0 under stop-and-go congested conditions when very few cars and trucks are able to maintain their FFS. The RMI would drop below one between these two extremes. A surrogate value for the numerator and denominator is the ratio of the average speed to FFS by vehicle class. A further extension of this concept is the ability to estimate a truck density equivalent for truck LOS estimation purposes with a known or estimated RMI. This estimation is accomplished using the following relationship: Equation 3Density Density RMI trucks pcars = The numerator in Equation 3 is taken as the computed HCM LOS measure for basic freeway segments, while the RMI in the denominator is assumed or estimated. The approach used in this research was to calibrate/develop speed prediction models for both trucks (three types) and passenger cars using a microscopic freeway simulation model (FREESIM). Other field based approaches were considered and discarded due to the complex- ity and cost of flow and speed data collection by vehicle class. The simulation model was first calibrated against sensor data extracted on the I-4 freeway in Orlando, FL. These data were supplemented with surveillance video data to retrieve actual counts of the various truck classes. Once the simulation model produced reasonable comparisons to the field data, it was used to develop statistical speed models for a representative basic freeway segment for four classes of vehicles (one passenger car and three truck categories) by varying factors for total volume (or volume per lane), percent trucks, road grade, and number of truck lane restrictions. Thus, each of the four speed models takes the form class intercept, volume, grade, percent trucks, restricted lanes, interaction terms Equation 4 S i f ( )( ) = The intercept term therefore represents the FFS for the subject vehicle class. Applying the models by class means the RMI, truck density, and LOS can be computed. This study explored the development of a method to assess LOS for trucks based on a maneu- verability measure, which was a function of relative percentages of FFS between trucks and pas- senger cars. There is some work needed to be done, such as revising the model to use a variable for the segment entering average speed of the vehicle class instead of using a base FFS. It would also be desirable to perform field verification of base FFS of trucks relative to passenger cars to serve as validation for the values that result from the simulation model. Washburn and Ko Washburn and Ko (2007; also see Ko, Washburn, and McLeod, 2009) conducted opinion surveys of 459 truck drivers and 38 carrier managers to identify the roadway, traffic, and other highway-related factors most important to them. The effects of truck regulations were explicitly excluded. They attempted to survey a cross section of driver types representative of different carrier and equipment types.

Literature Review 37 They found that drivers and carriers tended to place greater emphasis on different aspects of the highway experience. Drivers placed greatest importance on the quality of the ride and ease of driving (pavement smoothness, fewer maneuvers required, and ease of maneuvers). Carriers placed greatest emphasis on speed and travel time reliability. Based on the combined results from drivers and carriers, the authors recommended the fol- lowing key measures of LOS for evaluating truck performance on a facility: • Freeways—speed variance and pavement quality; • Two-lane highways—percent time being followed, percent time spent following, travel lane and shoulder widths, and pavement quality; and • Urban streets—ease of turning maneuvers, speed variance, traffic density, and pavement quality. Note that speed variance, in this case, refers to the “ease of maintaining a consistent speed” over the length of a trip and not the variation in average trip speed from trip to trip, which reflects to a certain extent the psychological comfort of a trip. It is essentially a measure of the amount and frequency of accelerations and decelerations during the trip. Even though motor vehicle carriers were surveyed, none of the recommended truck LOS measures dealt with speed and travel time reliability, which were the primary concerns of ship- pers and carriers. Instead, the recommended LOS measures of comfort and convenience focus on the needs and perspectives of truck drivers. 4.2 International Practice A scan of the international literature review found that several countries are still in the process of developing their highway capacity manuals. Significant differences in traffic laws of individual countries limit the transferability of procedures and PCEs adopted by other countries. The gen- eral finding from the literature scan is that most capacity analysis manuals from other coun- tries generally follow the HCM 2010 concept of converting trucks to PCEs and then computing capacity and performance for the equivalent passenger car stream. None of the international manuals reviewed to date provide performance measures or meth- odologies for measuring or predicting LOS from the point of view of truck shippers or carriers. Capacity analysis manuals and relevant research from the following countries were reviewed: • Germany, • The United Kingdom, • Canada, • China, • Indonesia, • Australia, • Brazil, • Japan, • India, • Thailand, and • Singapore. 4.2.1 Germany In Germany, heavy-vehicle percentages in traffic stream are used as parameters describing the influence of trucks on both freeways and rural highways. All vehicles with a maximum

38 incorporating truck Analysis into the Highway capacity Manual weight above 3.5 metric tons are considered as heavy vehicles (the maximum allowable weight is 40 metric tons, width is 2.6 m, and length is 18.75 m with some exceptions). Differences between German and U.S. traffic laws suggest that German PCEs and truck analy- sis methods may not be directly transferable to the United States. The maximum allowable speed for trucks in Germany is 80 km/h (49 mph), however, it is not strictly enforced. On the other hand, the maximum speed of passenger cars is 140 km/h (85 mph) on a level freeway. In general, prohibition of overtaking by trucks is implemented in German freeway networks. Right-hand overtaking by any vehicle of any other is prohibited in Germany, a rule that is generally obeyed. As a result, trucks and cars in freeways are segregated. These circumstances have resulted in trucks running on the far right lane nearly all the time. Trucks use the left or middle lane only for overtaking. Trucks are not allowed outside of the two right-hand lanes on a freeway. It is mandatory in Germany that trucks keep a distance of 50 m (150 ft) between them for safety. Occasionally, on some of the major freight-hauling freeways, long “freight-truck” queues are observed in the right lane. These moving truck queues create problems for car drivers exiting or entering the freeway at junctions. Currently, when computing performance measures, passenger car speed is the key input. In computing roadway performance measures, there is an implicit assumption that increasing traffic volumes adversely impact trucks and automobiles in the same manner and to the same degree. Road gradient, different speed-flow characteristics, and different capacities are also taken into account while analyzing performance measures.3 In 1994, the first draft of a German Highway Capacity Manual (German HCM) was presented on behalf of the Federal Minister of Transport to improve the practical applications of traffic engineering theories (Wu, 1998). The theoretical capacity on German motorways under ideal conditions (light and dry) is shown in Exhibit 9. Geistefeldt (2009) proposed a new empirical method for estimating PCEs for heavy vehicles on freeways. The proposed approach is based on the concept of stochastic capacities, illustrated by the capacity distribution functions. Capacity distribution functions were created using 5-min interval traffic counts from German freeways with varying geometric parameters. The empirical PCE estimation and the parameters of the corresponding capacity distribution functions vary from 1.3 to 2.6. The estimated PCEs tend to decrease with an increasing number of lanes. A comparative study was performed to compare saturation flow rate as presented in the German HCM. The study site was in the City of Dresden; the saturation flow rates with vary- ing grades and heavy-vehicle combinations were compared (Boltze, 2006). The basic capacity is 2,000 vehicles per hour. The influence of heavy vehicles and grade on the saturation flow is shown in Exhibit 10. 3 By way of comparison, the United States’ HCM 2010 defines a heavy vehicle as a vehicle with more than four wheels on the ground during normal operations. Truck percentage 6-lane motorways Metropolitan Long distance 0% 1820 2075 1815 5% 1780 2010 1790 10% 1730 1945 1765 15% 1690 1875 1740 *Capacities in vehicles per hour per lane; adapted from Wu, 1998. 4-lane motorways Exhibit 9. German motorway capacities under ideal conditions.*

Literature Review 39 Brilon and Bressler (2004) analyzed traffic flow characteristics on freeway upgrades in Germany. They used all external influences, degree of gradient, and length of upgrade together with traffic flow parameters such as volume or proportion of heavy vehicles through specific parameters. Based on this analysis, they found that capacity solely depends on the degree of gradient but is not influenced by gradient length. However, travel speed is significantly influenced by the degree of gradient and the length of the grade (up to L ≤ 4000 m) as well as by the proportion of trucks. 4.2.2 United Kingdom Traffic Capacity on Urban Roads (Department of Transport, Highways Agency, n.d.) provides capacity look-up tables for various types of roads according to the proportion (up to 15%) of heavy vehicles on the road. The recommended capacity adjustments for higher proportions (the heavy-vehicle percentage in a one way flow exceeds 15%) of heavy vehicles are shown in Exhibit 11. In the United Kingdom, the motorway speed limit is 60 mph or less within a built-up area. For urban all-purpose roads, the speed limit is either 40 mph or less for a single carriageway or 60 mph or less for a dual carriageway. 4.2.3 Canada The Canadian Capacity Guide for Signalized Intersections (Canadian Institute of Transporta- tion Engineers, 2008) incorporates heavy vehicles for design of traffic signals and analysis. Heavy vehicles are included as a passenger car unit equivalent, which is discussed in the following section. Passenger Car Unit Equivalent in Flow The Canadian Capacity Guide focuses on the movement of traffic flow units including trucks such as cars, transit vehicles, cyclists, and pedestrians at signalized intersections. Vehicular traffic flow is commonly expressed as a homogeneous entity by converting the individual vehicle class into passenger car units (PCUs). Three types of trucks are listed in the Canadian Capacity Guide. Exhibit 12 illustrates the truck classification and corresponding PCU. 0% Heavy Vehicles 10% HV 20% HV 30% HV Grade German HCM Dresden German HCM Dresden German HCM Dresden German HCM Dresden 0% 2000 2000 1860 1800 1540 1600 1380 1300 2.5% 1835 2000 1710 1800 1410 1600 1265 1300 3.0% 1800 1950 1680 1750 1385 1550 1240 1250 4.0% 1750 1850 1630 1600 1345 1450 1205 1200 5.0% 1700 1650 1585 1350 1310 1250 1170 1150 *Entries are saturation flow rates in vehicles per lane per hour of green for signalized intersections. German HCM is German Highway Capacity Manual. Source: Boltze, 2006. Exhibit 10. Dresden and German Hcm saturation flow rates.* Exhibit 11. Reduction in flow due to heavy vehicles. Heavy-vehicle % Total reduction in flow level (veh/h) UM and UAP dual carriageway road Single carriageway UAP road having width of 10 m or wider Single carriageway UAP road having width less than 10 m Per lane Per carriageway Per carriageway 15 20% 100 100 150 20 25% 150 150 225 Notes: UM = Urban motorway; UAP = Urban all-purpose road. Source: Department of Transport, Highways Agency, n.d.

40 incorporating truck Analysis into the Highway capacity Manual Saturation Flow Adjustment Factors The Canadian Capacity Guide suggests a number of adjustment factors to adjust the basic saturation flow values for heavy vehicles including other adjustment factors in the absence of directly measured saturation flows at the analyzed intersection. The adjusted saturation flow depends on the basic saturation flow and is a function of the applicable adjustment factors: Equation 5adj basic adjS S f F )(= where Sadj = adjusted saturation flow (pcu/h), Sbasic = basic saturation flow (pcu/h), f(Fadj) = adjustment functions, and Fadj = individual adjustment factors. Truck Size and Weight in Canada The truck size and weight regulations in the Canadian provinces in the 1960s were similar to those in the U.S. states. The detailed specifications were developed for tractor-semitrailers from 3 to 6 axles and A-, B-, and C-trains from 5 to 8 axles for interprovincial highway transporta- tion. Canadian provinces and territories have the authority to set, monitor, and enforce truck size and weight regulations. Woodrooffe et al. (2011) suggested that the process of implementa- tion was advancing slowly. The delay was due to public concern in Ontario with an increase in semitrailer length and overall length for doubles, which restricted full implementation in the six eastern provinces for five years. Researchers argued that there was national agreement among stakeholders that Canadian size and weight regulations were inconsistent and outdated, which contributed to cross country transport inefficiencies (Abdelgawad et al., 2010). Exclusive Truck Facilities in Canada Roorda et al. (2010) analyzed exclusive truck facilities in the Greater Toronto Area (GTA). The research was motivated from studies on truck facilities being conducted in many U.S. states (Florida, Texas, Virginia, etc.). Travel demand on the 400-series freeway is modeled and cali- brated in detail to reflect observed freeway traffic volume. Two scenarios were evaluated. In one scenario, the conversion of one lane in Highway 401 in each direction into an exclusive truck lane resulted in stable overall freeway demand for pas- senger cars and light truck trips and an increase in demand for medium and heavy truck trips by 5% to 15%. The scenario also resulted in reduced passenger car and light truck capacity and increased medium and heavy truck capacity on Highway 401. The resulting effect was approxi- mately stable freeway demand for passenger cars and light trucks and a significant increase in freeway demand for medium and heavy trucks over the base case. In the second scenario, construction of an exclusive truck highway in a hydro corridor across the GTA resulted in a 3% to 7% increase in passenger car/light truck trips on the freeway and an 8% to 13% increase in medium and heavy truck trips. Exhibit 12. canadian passenger car unit equivalents—signalized intersections. Truck types Passenger car unit equivalents (pcu/veh) Single unit trucks 1.5 Multi-unit trucks 2.5 Multi-unit trucks heavily loaded 3.5 Source: Canadian Institute of Transportation Engineers, 2008.

Literature Review 41 In the same study corridor, Abdelgawad et al. (2010) conducted a simulation study for exclu- sive truck lanes. Researchers evaluated two alternatives: in the first, addition of four-lane truck facilities resulted in greater travel time improvements for trucks, which resulted in the reduction of freeway average travel speeds in the network for both a.m. and p.m. peak hours. In the second alternative, conversion of a freeway lane to an exclusive truck lane on Highway 401 resulted in increased congestion for passenger cars, but improved travel speeds for trucks. Both of these scenarios show truck facility usage ranges from 100 to 800 trucks per hour per direction. 4.2.4 China A comprehensive highway capacity study was conducted from 1995 to 1999 with the purpose to develop draft capacity guidelines for roads and major intersections outside of urban areas (Bang and Heshen, 2000). Field data were collected at 144 road links and at 19 major inter- sections outside of urban areas. The project was intended to support central efforts towards the development of a complete Chinese Highway Capacity Manual. The following average traffic composition was recorded from the surveyed sites: • MC2—two-axle motorcycles 4%; • MV—mini-vehicles (3 and 4 axles) 10%; • LV—light vehicles (cars, vans, etc.) 27%; • MHV—medium heavy vehicles 25%; • LHV—large heavy vehicles 21%; • TC—truck combinations 8%; and • TRA—farm tractors 5%. The corresponding PCE for these types of vehicles is shown in Exhibit 13. The recommended base FFSs depend on road, terrain, and vehicle types and are shown in Exhibit 14. Exhibit 13. Passenger car equivalents for road links studied in china. Road type/ total both directions* Terrain type Traffic flow (veh/h) PCEs (PCE for LV = 1.0) MV MHV LHV TC TRA MC2 4/2 UD + D (CW = 13 16 m) Flat 0 1.3 1.4 1.6 2.2 3.2 0.5 2500 1.4 1.5 1.8 2.4 3.5 0.5 5000 1.2 1.2 1.4 2.0 2.5 0.3 Rolling 0 1.7 1.8 2.5 3.4 3.8 0.5 2100 1.9 2.0 2.5 3.4 4.2 0.5 4200 1.8 1.5 1.8 2.4 3.4 0.3 Hilly 0 1.8 2.0 3.1 4.2 4.4 0.3 1750 2.0 2.3 3.1 4.2 4.9 0.4 3500 1.9 1.7 2.4 3.4 3.9 0.3 *Notes: UD = undivided; D = divided; CW = carriageway (roadway width). Exhibit 14. Base free-flow speed for interurban and township road in china. Road type/ both directions Terrain type Base free-flow speed (km/h) LV MV MHV LHV TC TRA Motorway Flat 90 70 70 65 60 — Rolling/ hilly 80 60 60 52 50 — Multilane road >13 m Flat 70 55 62 62 54 25 Rolling 65 50 57 55 47 23 Hilly 60 45 51 48 39 20 Source: Bang and Heshen, 2000.

42 incorporating truck Analysis into the Highway capacity Manual 4.2.5 Indonesia The Indonesian highway agency realized that the existing capacity manuals from developed countries could not be successfully implemented in Indonesia because Indonesian traffic charac- teristics differ from those of developed countries (Indonesian Directorate General of Highways, 1993). Thus, data collection was performed at a total of 147 sites in 16 cities in Indonesia to develop capacity parameters appropriate for Indonesia. Based on this data collection effort, the following PCEs were identified. For signalized inter- section analysis, a heavy-vehicle factor of 1.3 is used to convert to PCUs. For urban roads, the default PCU value of 1.3 is used for heavy vehicles. If there are a lot of heavy vehicles, a PCU of 2.0 could be used. Heavy vehicles are classified as buses, two-axle trucks, three-axle trucks, and truck combinations. Heavy vehicles classifications include the following (Indonesian Directorate General of High- ways, 1995): • Medium heavy vehicle (MHV): two-axle motor vehicles with an axle spacing of 3.5 to 5.0 m, including buses and two-axle trucks with six wheels; • Large trucks (LT): three-axle trucks and truck combinations with axle spacing from first to second axle of <3.5 m; and • Large bus (LB): two- or three-axle buses with an axle spacing 5.0 to 6.0 m. For interurban roads and motorways, capacity is measured in light vehicle units (LVUs). Two sets of LVU values are used with different criteria for equivalency (see Exhibit 15): • Speed-based LVU values are based on the relative impact on light vehicle speed due to differ- ent types of vehicles in the traffic stream; and • Capacity-based LVU values are based on the relative impact on capacity due to different vehi- cle types. The Indonesian Highway Capacity Manual suggests that the FFS for a passenger car is typically 10% to 15% higher than that for other types of light vehicles. The actual capacity is adjusted from ideal capacity by incorporating a road width adjustment factor, a directional split adjustment factor, a motorcycle traffic adjustment factor, and a side friction adjustment factor. The calcula- tion procedures given in the manual are in some cases similar to the U.S. HCM; users are advised to use values for Indonesian conditions as appropriate. For motorways, the Indonesian HCM recommends a FFS of 85 km/h (52 mph) and a base capacity of 2,300 LVUs/h/l, respectively, for a four-lane divided motorway in flat terrain. Indonesia does not use the U.S. HCM LOS concept; therefore, speed and degree of satura- tion are used in the Indonesian HCM. Speeds are much lower in Indonesia than in the United States for a given degree of saturation (flow/capacity = Q/C) (Indonesian Directorate General of Highways, 1993). Exhibit 15. Light vehicle units conversion in Indonesian Hcm. Terrain/Road Type LVU (speed) LVU (capacity) MHV LB LT MHV LB LT Flat terrain/Divided road 1.5 1.0 3.2 1.2 1.5 2.0 Flat terrain/Undivided road 1.5 1.2 2.7 1.2 1.5 2.0 Rolling terrain/All road types 2.0 1.3 4.0 1.3 1.7 2.5 Hilly terrain/All road types 3.5 1.5 5.5 1.5 2.0 3.0 Source: Indonesian Directorate General of Highways, 1995.

Literature Review 43 4.2.6 Australia Austroads published Guide to Traffic Management for traffic studies and analysis. The guide provides guidance on traffic analysis for uninterrupted and interrupted flow for various types of intersections. Different factors affecting capacity and LOS due to roadway condition, traffic composition, and so forth are also presented in the guide (Austroads, n.d.). The document defines “truck” as a vehicle with more than four single tires and involved primarily in the transport of goods and services. It utilizes the HCM 2000 methodologies for calculating capacity, delay, and LOS on transportation facilities. A suggestion is made in the document that when using the HCM 2000 procedures, the vehicle equivalency factors should be adjusted to reflect the characteristics of Australian trucks. However, there is no evidence in the document that provides guidelines on appropriate values for vehicle equivalency of Australian trucks that should be used in the analysis. 4.2.7 Brazil A study was conducted in Brazil to estimate truck PCEs for divided multilane highways (Cunha and Setti, 2011). In Brazil, trucks represent a high proportion of highway traffic and they are longer, heavier, and have smaller engines than the trucks used in the development of the HCM 2000. Truck characteristics (power, weight, etc.) were observed at several weigh stations on multilane highways. A microscopic traffic simulation software, CORSIM’s heavy-vehicle performance and car-following models were recalibrated using a genetic algorithm with truck performance data and traffic data collected on a divided multilane highway. The recalibrated CORSIM was then used to derive new PCEs. PCE tables for specific grades and for extended segments were created to replace those used in the HCM 2000. The results show the need for development of a Brazilian HCM. They suggested that the use of the PCEs found in this study may improve LOS estimates rather than to adapt from the HCM to Brazil. Demarchi and Setti (2003) used two types of trucks in the analysis for illustration purposes with varying mass-to- power and lengths. The results indicate that the errors in the estimation of equivalent flow rates are negligible for densities less than 10 veh/(km-lane), but increase significantly with the increase in density. The derivation of an aggregate PCE could avoid this problem. 4.2.8 Japan The latest trucking research in Japan is concerned with futuristic automated truck lanes, which promise to reduce congestion and increase safety. For purposes of analyzing such lanes, Morikawa, Miwa, and Sun (2011) studied the New Tomei Expressway and obtained—through maximum likelihood estimation—a PCE value of 1.73. Rahman, Okura, and Nakamura (2003) suggested a method for estimating PCE in Japan for large vehicles at signalized intersections based on increased delay caused by the large vehicles. Researchers found that for the same percentage of heavy vehicles, the PCE value varies con- siderably with the position of large vehicles in the queue. In this study, a queue length of 8 to 17 vehicles was used to develop PCE values at signalized intersections. 4.2.9 India The Central Road Research Institute (Indian Central Road Research Institute, 1988) adopted a linear regression analysis technique for determining PCE for different classes of vehicles, including trucks. Aggarwal (2011) developed a fuzzy based model for the estimation of PCE value for trucks based on inputs such as pavement width, shoulder condition, directional split,

44 incorporating truck Analysis into the Highway capacity Manual and speed of the traffic. Most of the research has provided significant insight about mixed traffic operation in India, but has recommended only static PCE values for trucks and other vehicle categories for different roadways and control conditions. 4.2.10 Thailand The wide variety of vehicles in Thai roads requires a comprehensive approach to PCE. Exhibit 16 is an adaptation of a table published by Mathetharan (1997), which is still used in Thailand to homogenize traffic streams. Minh and Sano (2003) studied the influence of motorcycles on saturation flow rates in Hanoi and Bangkok and arrived at a PCE of 0.24 and 0.18, respectively. While motorcycle PCEs may not be of interest for this project, the methods used in this project—plotting the proportion of motorcycles against the saturation flow rate—appear to be representative of Thai PCE research. Another study conducted in Thailand suggested that the overall effect on the capacity with the prevailing proportion of large-sized vehicles resulted in reduction in capacity on the order of 15%. PCEs for medium- and large-sized vehicles are obtained as 1.0 and 1.5 respectively (Tanaboriboon and Aryal, 1990). 4.2.11 Singapore A comprehensive 2-year study of truck traffic at 219 Singaporean sites was conducted by Fwa, Ang, and Goh (1996). It was found that the time distribution of truck travel varies greatly among the five roadway classes (i.e., expressways, arterials, collectors, industrial roads, and local roads). In a similar study, Fan (1990) suggested PCE values of 1.3, 2.6, and 2.7 for light trucks, heavy trucks, and buses, respectively, for Singapore expressways. These PCE values are higher than those recommended for use in the United States. 4.3 Conclusions of Literature Review Review of international literature and practice found that most countries use PCEs like the U.S. HCM to convert trucks in the traffic stream into the equivalent number of passenger cars before computing capacity and speed. Unlike the U.S. HCM—which uses a single class Vehicle PCE Motorcycle 0.25 Passenger car 1.00 Taxi 4-wheel 1.00 Tuk-Tuk (3-wheel) 0.75 Bus Light 1.25 Medium 1.50 Heavy 2.00 Truck 4-wheel 1.75 6-wheel 1.75 10-wheel 2.00 Articulated 3.00 Source: Mathetharan, 1997. Exhibit 16. Thailand PcEs.

Literature Review 45 of trucks—China, Indonesia, Singapore, Thailand, and Canada subdivide trucks into three or four subtypes. U.S. research suggests that truck PCEs used to compute saturation flow rates at signalized intersections should vary by truck size (i.e., the number of axles). The U.S. HCM currently uses just a single truck class (plus separate classes for buses and RVs). Japanese research found that the PCE effect of a truck on saturation flow rates also varies by the position of the truck in the signal queue. Similar findings were reported in U.S. research (Washburn and Cruz-Casas, 2010). Chinese, Brazilian, Canadian, Indonesian, and U.S. research all confirm that PCEs vary by weight to equivalent horsepower ratio (WEHPR). U.S. research confirms the HCM PCE tables, which show that the effects of trucks decrease as trucks make up a larger proportion of traffic stream. This same research, however, suggests that HCM tables should be extended above 25% trucks in the traffic stream. U.S. research suggests that vehicle length (as opposed to WEHPR) affects truck PCEs for freeways on level sections. German research has identified differing effects of different percent of heavy vehicles on signal saturation flow rates as a function of approach grade. For freeways, German research has found that the PCEs of heavy vehicles decrease with increasing number of lanes. Indian research found that PCEs increased 20% over a 14-year period in that country, perhaps due to the evolution of the vehicle mix and vehicle WEHPRs, which suggests that the U.S. HCM PCE tables should be specifically tied to a specific distribution of truck subtypes (WEHPRs) and updated regularly. This literature research on international truck analysis did not identify any LOS analysis per- formance measures or procedures designed to specifically represent the perspectives of truck shippers or carriers. All international capacity manuals do provide PCE parameters to convert trucks to equivalent PCUs before estimating the equivalent capacity and speed for the passenger equivalent traffic stream. The differences in traffic laws among countries and the horsepower to weight ratios of trucks in other countries suggest that actual PCE values cannot be transferred directly to U.S. practice. Several countries provide additional gradations of truck types beyond the simplistic U.S. HCM method of truck, bus, and RV. These classification schemes may provide some useful ideas for an augmented U.S. HCM truck classification scheme.