Considerable debate has focused on the safety of commercial heavy trucks, and particularly questioning if changing truck sizes and weights would alter roadway safety. As noted in the Comprehensive Truck Size and Weight (CTS&W) Study, the safety of freight moving on the roadway is a combination of many factors: vehicle miles traveled (VMT) or exposure; vehicle performance characteristics; driver performance and ability; enforcement; roadway design; road conditions; motor carrier management; and vehicle condition and maintenance. Among these factors isolating the impact of TS&W is difficult. Because larger and heavier trucks are a relatively small subgroup of all trucks, differentiating their crash involvement patterns from those of other truck types is problematic. This study will discuss the safety performance and exposure factors for the Western Uniformity Scenario vehicles.
Discussing these safety aspects is not intended to diminish other facets of the safety picture. Difficult to quantify aspects, such as alternative enforcement mechanisms, vehicle maintenance and driver qualifications are discussed in the CTS&W Study.
The analysis presented in this chapter confirms three important factors highlighted in the CTS&W Study. First, travel on undivided, higher speed-limit roads with many at-grade intersections and entrances significantly increases crash risks compared to travel on Interstate and other roadways with design characteristics similar to Interstate highways. Second, higher traffic density increases the crash risk. Third, TS&W policies can influence vehicle stability and control because they directly impact key vehicle design attributes such as number of axles, track width, wheelbase, number of units in a combination, loaded weight and overall length.
This chapter contains two major sections. The first highlights the vehicle safety performance analysis undertaken for Western Uniformity Scenario. The analysis uses the same methodology as the CTS&W Study to examine current western LCVs and the Scenario's LCVs. The second section reviews recent crash data analysis and presents an updated crash data analysis.
Three performance measures are often used as indicators of a truck's crash risk: Static Rollover Threshold, Rearward Amplification, and the Load Transfer Ratio. All three metrics describe aspects of a vehicle's inherent propensity to rollover. Crashes where the first event was a truck rollover accounted for 20 percent of the fatal single-vehicle crashes for large trucks in 2000.22
Both the current population of LCVs operating in the Western States and the Western Uniformity vehicles were analyzed. All the Scenario's vehicles currently operate in some part of the study region. This allowed the researcher to obtain “real-world” physical measurements for the stability input variables (for example, king pin setting, axle spacings, typical loads etc.).
In addition, current LCVs were analyzed because the large differences in State LCV regulations produce significant variations in vehicle design. These differences also translate into variations in vehicle dynamic performance. A sizable effort was made to fully understand the priorities and constraints unique to the Western States that would influence vehicle design particularly with respect to safety performance optimizations. The simulation and safety analysis reflects basic vehicle design, commodity types and loading variations occurring in the Western States.
Table VII-1 shows the configuration, body type, trailer length(s), articulation type and GVW for the analyzed vehicles. These parameters were used to determine the vehicle stability and control performance.
The analysis includes van-type configurations including: the 5-axle tractor semitrailer, STAA Doubles, Rocky Mountains Doubles (RMD), Turnpike Doubles (TPD) and Triples. The analysis shows results for both the A-train and C-train Triples configurations. The tank vehicles are truck trailers, RMD A-train and its equivalent B-train configuration. This Chapter provides the results for the 96-inch axle width.23
The articulation type (see Figure VII-3) strongly influences the vehicle's stability and control. Both the C-dolly and B-train connections effectively eliminate an articulation point and increase stability and dynamic control of the vehicle. On the other hand, the reduced articulation decreases the maneuverability of the vehicle through curves and turns, impacting roadway geometry and traffic operations. The same trade-off exists for the vehicles with wider axle widths (see Woodrooffe, 2003).
The vehicles are evaluated at their respective maximum allowable weight conditions since the Western LCVs used for bulk transport are typically loaded to their maximum allowable gross weight. This represents the most severe operational case for vehicle stability since the center of gravity is the highest.
|Configuration||Body Type||Trailer Length (Feet)||Articulation Type||GVW (Pounds)|
|STAA Double||Van||28 X 28||A-Train||80,000|
|Turnpike Double||Van||45 X 45||A-Train||129,000|
|Turnpike Double||Van||48 X 48||A-Train||129,000|
|Rocky Mountain Double||Van||48 X 28||A-Train||117,000|
|Rocky Mountain Double||Van||48 X 28||A-Train||113,000|
|Rocky Mountain Double||Hopper||35 X20||A-Train||105,500|
|Rocky Mountain Double||Hopper||38 X 28||A-Train||123,000|
|Rocky Mountain Double||Hopper||38 x 20||A-Train||109,250|
|Rocky Mountain Double||Hopper||38 x 27||A-Train||127,500|
|Rocky Mountain Double (long)||Hopper||44 x 31||A-Train||123,000|
|Rocky Mountain Double||Hopper||38 x 27||A-Train||128,000|
|Rock Mountain Double||Tank||41 X 22||A-Train||117,000|
|Rock Mountain Double||Tank||41 X 22||B-Train||117,000|
|Triple||Van||28 X 28 X 28||A-Train||118,500|
|Triple||Van||28 X 28 X 28||C-Train||118,500|
The load for all van type trailers is 'general freight' occupying the full height of the load space. This assumes general freight has 30 percent of the payload in the top half of the container and 70 percent in the bottom half. Thus, the load has a relatively high center of gravity but not as high as a homogeneous load of the same dimensions. The analysis for the CTS&W Study used a homogeneous load to maximize the instability. Such a loading is atypical. Fancher et al24 found general freight to have the 30 - 70 distribution and that is the distribution used in the present analysis.
This section discusses the stability results for Static Rollover Threshold, Rearward Amplification and Load Transfer Ratio. These factors measure the impact of a sudden lane change, swerve and/or curve if negotiated at too great a speed potentially resulting in a rollover accident.
Static Rollover Threshold is a significant vehicle performance measure because it reflects overall vehicle stability for both emergency lane changes and typical negotiation of a well-designed roadway curve. The likelihood for a vehicle having to perform an evasive maneuver during a given trip is very low. By comparison, all vehicles must routinely negotiate curves and turns. In either case, vehicles with a low Static Rollover Threshold will be challenged.
The Static Rollover Threshold is the minimum amount of lateral acceleration needed to result in wheel lift-off from the ground - the point at which the vehicle then rolls over (Figure VII-1). Higher scores indicate better performance. Larger, heavier vehicles do not necessarily have poorer Static Rollover Threshold performance than smaller, lighter ones. The important variables are how the payload is distributed along the length of the vehicle and the height of the center of gravity. In general the lower the center of gravity and the more uniformly distributed the payload then the more stable the vehicle.
Illustration of Rollover Initiation
Static Roll Threshold can vary with the density of the commodity by up to 25 percent. On the other hand, since each van trailer is “loaded” with 70 percent of the load in the bottom half and 30 percent in the top half, all van trailers will have a rollover threshold of approximately 0.36g. These values represent a van with 96-inch width. Most new van trailers have a 102-inch width, producing a 5 percent improvement in roll stability. Static Rollover Threshold of the tanker fleet is approximately 0.40g, that is approximately 22 percent better than van trailers. This is due to the lower center of mass of the tank trailers relative to the van trailers.
Figure VII-2 shows the Static Rollover Threshold for 17 analyzed vehicles. The first 8 were chosen from the fleet of current vehicles operating in the West and the latter 9 are the Scenario vehicles - some of which already operate in the West. Static Rollover Threshold less than 0.30 is considered very poor, between 0.30 and 0.35 is poor, between 0.35 and 0.40 is good and greater than 0.40 is excellent. All the configurations analyzed have a good to excellent rating for Static Rollover Threshold.
As shown in Figure VII-2, most of the van trailer Scenario vehicles perform worse than the STAA double. This is critical since currently over 50 percent of the LCVs in the Western States are van-trailers,25 a pattern that would continue under the Scenario. Past studies have shown that the Static Rollover Threshold can be improved through different vehicle designs - such as wider vehicles, lower floor heights; new equipment such as enhanced electronic braking, tire and suspension systems; and B-train and C-dolly trailer connections (see Figure VII-3). The B-train improvement can be seen comparing the final two vehicles in Figure VII-2, a RMD tanker versus the B-train, with all other variables held constant, there is a 3 percent improvement.
Static Rollover Threshold
Current Vehicles and Scenario Vehicles
When articulated vehicles undergo rapid steering, the effect at the trailer is magnified. This can result in excessive movement of trailers, which can be very dangerous if, for instance, they move into other lanes and interfere with other vehicles. In the extreme, LCVs' rearward amplification can cause rear trailers to rollover.
Rearward amplification is influenced by the center-of-gravity, axle group weights, wheelbase dimensions, coupling types and locations, drawbar dimensions, suspension and tire characteristics. Mathematically, rearward amplification is the ratio of the lateral acceleration experienced at the rearmost trailer in a combination to that of the tractor, when a lane-change evasive maneuver is executed. In this case, values of 2.0 or less indicate acceptable performance.26
Figure VII-4 shows that certain configurations of LCVs are more prone than typical tractor-semitrailers to rearward amplification. For example, rearward amplification is 1.73 times greater for trucks with twin 28-foot trailers than for typical tractor semi-trailers, but can be 2.18 times worse for triple-trailers (with their five points of articulation) than it is for typical tractor semi-trailers. The type of mechanism connecting the trailers can affect the rearward amplification. A mechanism with a single connection point to the lead trailer, the most common type in the United States, is referred to as an “A-train” or “A” converter dolly. This type allows more rearward amplification than does the “B-train” or “C-train”. Figure VII-3 illustrates and discusses these different connecting mechanisms.
Figure VII-4 shows that typical tractor-semitrailer combinations have a rearward amplification of 1.24. Currently -- designed STAA doubles (two 28-foot trailers) have rearward amplifications of 2.15. Also the RMD at 105,500 pounds with medium and light density commodities would be considered to have poor dynamic performance. The Triple-trailer A-train, at a value of 2.72, has the highest Rearward Amplification of all vehicles examined. When the Triple-trailer combination is fitted with C-dollies, dynamic activity is reduced by 39 percent and is in line with the remaining vehicles. The most stable vehicle examined is the B-train tanker at 117,000 pounds and a trailer width of 102" - it performs even better than the single-trailer combination.
In the case of multi-trailer combinations, roll coupling is a vehicle design feature that counters dynamic roll instability. It uses a coupling feature designed to take advantage of the fact that two adjacent units in a multi-trailer combination roll in different directions during a dynamic lane change maneuver. By making the coupling or hitch more rigid along the roll axis, each unit in the combination "helps" the other counteract excessive roll forces.
Roll coupling is a special attribute of "B-train" and "C-dolly" connections. A "B-train" connection between two trailers in a twin configuration essentially creates a semitrailer/semitrailer combination with two articulation points instead of three. A standard "fifth-wheel" connection is used to couple the two trailers together, thereby providing significant counter-roll forces between the two trailers.
A "C-dolly" connection also provides roll and coupling stiffness through the use of two drawbars between trailers. "A-dollies", which are used today, have one drawbar. Both B-train and C-dolly connections between two trailers effectively eliminate an articulation point and provide a large counter-roll force for each of the two trailers when dynamic forces act in opposing directions during an evasive lane change maneuver.
Some researchers believe the same effect can be accomplished through the use of such advanced technology as electronically controlled braking systems (currently the subject of a field operational test), which employ load and speed sensitive differential braking to maintain the direction of the individual units in combination vehicles making evasive maneuvers. This could reduce the crack-the-whip phenomenon and dynamic roll instability especially inherent in multi-trailer vehicles.
Rearward Amplification: Current and Scenario Vehicles
Load transfer ratio is the proportion of load transferred to one side of a vehicle in a transient evasive steering maneuver. When the load transfer ratio reaches 1, the entire vertical component of the load is being transferred through the wheels on one side of the vehicle and rollover is about to occur. The load transfer ratio directly expresses the proximity to rollover in rapid maneuvers and emergency avoidance situations. In important respects, the load transfer ratio combines the influence of steady-state rollover and rearward amplification in one performance index. The load transfer ratio is computed for a standard maneuver based on a standard steering input. Load transfer ratio is influenced by the center-of-gravity, axle group weights, vehicle width, lateral load shift, wheelbase dimensions, coupling types and locations, drawbar dimensions, suspension characteristics and tire characteristics.
The Canadian performance standards recommend when a loaded vehicle "negotiates an obstacle avoidance, or lane change maneuver at highway speeds, the load transfer ratio should not exceed 0.60." 27 This is the generally accepted standard for other jurisdictions that employ performance measures. Of the current double configurations, only the STAA double and 105,500 pound LCV were found to have sub-standard performance.
Among the Scenario vehicles, Figure VII-6 shows that the B-train fuel tanker has the most stable characteristics followed closely by the Triple C-train. The improved performance by the B- and C-train configuration is attributed to the elimination of one articulation point per trailer and the addition of roll coupling between trailers. The Load Transfer Ratio of the RMD fuel tanker with a GVW 117,000 pounds was 2.4 times greater than the B-train tanker at the same GVW. However, the RMD compared favorably with the other vehicle classes including the tractor semi-trailer. This underscores the superior characteristics of the B- and C-train configurations.
The Load Transfer Ratio performance of the Triple A-train and the STAA Doubles is very poor. In the simulation, the Triple A-train achieved the theoretical maximum of unity, which means that the vehicle would have rolled over given the standardized test maneuver.
The TPD at GVW 129,000 pounds with 48-foot trailers out-performed the tractor semi-trailer at a GVW 80,000 pounds. When the trailers of the TPD were shortened from 48 feet to 45 feet, the Load Transfer Ratio increased by approximately 7 percent. This finding indicates that the stability performance of the TPD improves with trailer length.
Load Transfer Ratio Current and Scenario Vehicles
Many studies have attempted to identify how crash propensity varies with TS&W, with particular focus on doubles and/or LCVs. Some studies have reported that multiple-trailer trucks have lower crash rates than single-trailer trucks and other studies have reported the opposite. The disparity in findings is explained, in large part, by the difficulty in analyzing a relatively small population of vehicles and obtaining reliable accurate VMT and crash data for each vehicle type. To try to overcome these difficulties researchers have used various methodologies and data sets in different studies, resulting in different conclusions.
Prior studies are examined in an attempt to obtain indications of potential crash impacts associated with estimated changes under the Western Uniformity Scenario. Past studies do differentiate between single-trailer combination trucks and multi-trailer combination trucks, but the multi-trailer group includes STAA doubles (tractor and two 28-foot trailers) along with RMDs, TPDs, and Triples. STAA doubles dominate the multi-trailer crash history since they are the most common multi-trailer combinations. Nevertheless, the multi-trailer/single-trailer distinction is still important since LCVs are configured similar to STAA doubles and have similar dynamic handling/stability performance characteristics. But as discussed in the previous section on stability and control, most longer LCVs are more stable than STAA doubles although they off-track to a much greater degree. The greater length of LCVs may also make passing maneuvers on two-lane highways, and merging and weaving maneuvers on freeways more difficult than for shorter STAA doubles.
There have been attempts to isolate LCVs' crash experience from STAA doubles through surveys, making assumptions about past and future operating environments, or analyzing the data from countries that allow wider use of LCVs. All these efforts have fallen under criticism. Studies using surveys have difficulty matching the survey respondents to the VMT estimates, or the sample set is not large enough, or the sampled population contains a self-selection bias. Several studies assume that relaxed LCV regulations would translate into LCVs operating on roadways similar to current single trailer combination trucks, even though most States that allow LCVs limit their use to certain highways. Although these State-permitted highways include some two-lane highways, all States that allow LCVs restrict them to a subset of the highway network that is available to single-trailer combination trucks. The assumption that LCVs could operate throughout a State's highway network increases estimated multi-trailer fatal crash rates, since fatal crash rates are higher on rural non-Interstate highways. Finally, there have been attempts to utilize Canadian or Australian data to predict the safety impacts associated with more widespread LCV use in the U.S., but those countries have very different enforcement mechanisms, road networks, and traffic densities, making it difficult to draw implications from their crash experience for the U.S.
The following section discusses the methodology, data, and results for seven recent statistical examinations of multi-trailer combination vehicle safety. Table VII-7 summarizes results from those studies applicable to the current Scenario. Although several of the studies present data on single-unit trucks, this summary does not include that information since the Scenario would not impact those vehicles.
The General Accounting Office (GAO) reviewed nine LCV safety studies from 1978 through 1991 and found that they have disparate conclusions, including findings that multi-trailer vehicles are both more and less likely to be involved in crashes than other commercial vehicles. They found the reasons for the opposing conclusions rest with the different approaches used by the researchers, and the difficulty of collecting and interpreting the data used in the studies.
The GAO highlighted the problem of determining the safety of LCVs using crash data predominately based on the experience of twin 28-foot trailers. The GAO recommended an improvement of "truck accident and travel data, especially as they relate to the reporting of nonfatal accidents, the estimates of truck travel, and the identification of truck configurations."
The purpose of this study was to compare the crash experience of single-trailer combination trucks and the twin 28-foot STAA double following the 1982 Surface Transportation Assistance Act (STAA) in which States were required to allow twin 28-foot trailers on the National Network for Large Trucks (NN). Despite the study's title, it only analyzes the STAA double and not any other larger dimensioned vehicles. Thirteen States provided detailed truck data but only 4 States (Iowa, Kansas, Missouri, and Utah) provided data over the entire collection period from 1983 - 1991.
The study found that twin-trailer trucks had a lower fatal involvement rate than single trailer trucks given their current distribution of travel by functional class, but predicted a rate similar to single trailer trucks would result if they had the same distribution of travel.
This study focuses on the fatal crash involvement for single- and double-trailer combinations. Data on fatal truck crashes (1980 - 1984) and truck travel (1985) were used to estimate the effects of truck configurations, road class and operating environment on crash rates. Researchers reviewed the crash and travel data, conducting follow-up interviews to fill in gaps. The travel data was obtained from telephone surveys of about 5,000 trucks, sampled from the 1983 R.L. Polk vehicle registration file. The fatal crash data was from Trucks Involved In Fatal Accidents (TIFA). Table VII-2 shows the crash and travel data for two vehicle classes.
|Configuration||1985 VMT (millions)||1980 - 1984
100 million VMT)1
1. According to the National Highway Traffic Safety Administration's Fatal Accident Reporting System, the average number of fatal involvements by all heavy trucks was 4,294 per year during 1980 – 1984 period and 4,492 in 1985. To obtain 1985 fatal involvement rates for each of the three types of truck configurations, 4492/4294 multiplied average 1980 – 1984 fatal involvements.
Table VII-3 shows estimates of the fatal crash rates by vehicle group for the different road environments derived from data in the 1988 UMTRI study. This analysis highlights the importance of a truck's operating environment. Regardless of the number of trailers, limited access roads have much lower fatal involvements relative to the travel on those roads than do uncontrolled access highways. An examination of the single-trailer fatal crash record shows that 39 percent of the single-trailer VMT is on rural limited access but only 14 percent of the single-trailer fatalities occur on these roads. In this UMTRI study twin-trailer combinations were found to have somewhat lower crash rates than single-trailer combinations on all highways types except other rural highways on which their crash rates were substantially higher. Because twin trailers travel relatively less on other rural highways than single trailer combinations, their overall crash rates were found to be lower than those of single-trailer combinations. When researchers adjusted crash rates by assuming that travel characteristics by highway type would be the same for multi- and single-trailer combinations, the overall crash rate for multi-trailer combinations increased from 8.6 to 11.2 involvements per million VMT, a rate higher than the 10.2 involvements per million VMT for single-trailer combinations. These results highlight the issue that changes in fatal crashes will depend on the truck's operating environment.
|Configuration||1985 VMT (millions)||1980 - 1984
100 million VMT)
|Limited Access Rural||12,891||2,775||4.50|
|Limited Access Urban||6,602||1,829||5.80|
|Limited Access Rural||886||172||4.06|
|Limited Access Urban||569||117||4.30|
Computed using Campbell et al Appendix Table 15 Normalized Fatal Accident Involvement Rates by 8 Travel Categories for 5 Truck Types or Configurations NTTIS and 1980 - 1984 TIFA Files.
According to the National Highway Traffic Safety Administration's Fatal Accident Reporting System, the average number of fatal involvements by all heavy trucks was 4,294 per year during the 1980 - 1984 period and 4,492 in 1985. To obtain 1985 fatal involvement rates for each of the truck configurations, average 1980 - 1984 fatal involvements were multiplied by 4,492/4,294.
Limited Access includes Interstate and Other Freeways and Expressways.
The Transportation Research Board (TRB) utilized the fatal crash rates from Campbell et. al. (1988) adjusted assuming the same distribution of mileage by highway type for each type of vehicle. This assumed that double-trailer distribution of travel would become the same as the current usage of single-trailer tractor-semitrailers. TRB expanded the fatal crash rates using estimates of injury and property-damage-only crashes (Tables VII-4 and VII-5). In the TRB report there was no definitive statement regarding the relative safety of alternative LCV configurations. They did discuss the impact of increasing weights on a given vehicle, but many conclusions are difficult to interpret for this study because findings for combinations and straight trucks are grouped. For all combination vehicles TRB presented data showing that crash rates on limited access highways increase slightly up to a gross vehicle weight of about 60,000 pounds after which they generally level off. On other types of highways crash rates for combinations continue to increase up to about 75,000 pounds, but drop somewhat at higher weights.
|Crash||Severity Class ( percent)|
|Property damage only||70|
|Vehicle||Type of Crash (per 100 million VMT)|
|Fatal||Injury||Property Damage Only|
|Single Unit Trucks||7.7||185||499|
This study focused on estimating fatal crash rates for a single year using large federal databases containing crash, fatalities, and travel data. The crash data are from Fatal Accident Reporting System (FARS) and Trucks Involved in Fatal Accidents Program (TIFA). The researchers derived the vehicle travel estimates using FHWA's State reported travel data and the Truck Inventory and Use Survey (TIUS, predecessor to the current Vehicle Inventory and Use Survey (VIUS)).
The study produced a multitude of crash rates for different years using different data sources. Using FARS and adjusted FHWA travel data the authors estimated the 1988 fatal crash rates for tractor semi-trailers was 4.8 per 100 million VMT and for double-trailer combinations was 5.9 per 100 million VMT. Using TIFA and adjusted FHWA travel data the authors estimated the 1986 fatal crash rates were 4.33 per 100 million VMT for single-trailer combinations and 6.35 per 100 million VMT for multi-trailer combinations. Using TIFA and TIUS they estimated the 1986 fatal crash rates were 6.0 per 100 million VMT for single-trailers and 9.9 per 100 million VMT for multi-trailers. The authors stated that the TIFA and TIUS estimate is the best since TIUS is a sample.
This study reviewed the operations and crash rates for LCVs in Alberta, Canada for 1995 through 1998. The Canadian LCVs are similar to LCVs operating in the northwestern United States such as Montana and North Dakota. In Alberta a RMD consists of a 40 to 53 feet semi-trailer and a shorter 24 to 28 feet semi-trailer; a TPD consists of two trailers where both are between 40 and 53 feet; and a Triple trailer combination consists of three trailers all between 24 and 28 feet. Alberta requires selective routing, restrictions on vehicle speed, restricted time of day operation, enhanced driver qualification requirements and operating restrictions for adverse road and weather conditions. In general the operating network is restricted to multi-lane highways with four or more driving lanes except RMDs that may travel on a few two-lane highways.
During the study period there were 53 LCV crashes of which two were fatalities. The crash rates focus on the 37 rural crashes, but not on the 16 urban collisions due to difficulties computing the vehicle kilometers traveled in urban areas. This introduces a bias since the study only analyzes the best performing roads (4 lane rural). Table VII-6 shows the collision rates per 100 million kilometers and 100 million miles.
The study period contains only two fatalities attributed to TPDs implying a doubles crash fatal crash rate of 0.88 per 100 million kilometers or 1.42 per 100 million miles.
|Vehicle Type||Crash||Total Distance Traveled (100 million Km)||Total Distance Traveled (100 million miles)||Crash Rate (per 100 million Km)||Crash Rate (per 100 million Miles)|
|Rocky Mountain Doubles||11||1.07||0.66||10.28||16.55|
|LCV Doubles - all||31||2.26||1.40||13.72||22.09|
*Crashes for the LCV sub network only - no urban miles included
To convert to miles, kilometers was multiplied by 0.621.
Multi-Trailer includes RMDs, TPDs and Triples
Seventy-five commercial motor carriers participated in the study comparing crash rates of LCVs to Non-LCVs. All participants operated both LCV's and Non-LCVs. Crash and exposure data covered 1989 - 1994. This study focused on crashes that required the filing of a police crash report, an insurance crash report or recording of information in the motor carrier's crash register.
Among study participants, the mean crash rate was 887.25 crashes per million VMT for LCV's versus 1786.45 crashes per 100 million VMT for Non-LCVs. The difference in the mean crash rates was found to be statistically significant. The fatal crash rate for single-trailers was 24 per 100 million VMT while the LCV rate was 21 per 100 million VMT for the carriers in their study. Even though the crash rate was lower for LCVs, the researchers found that LCV crashes are more severe than non-LCVs: "the average number of fatalities per LCV crash was 90 percent higher than for each non-LCV crash."
The researchers discussed the possible foundations for the crash rate differential noting that LCV operators in their study predominately operated in rural areas on higher quality roads, possessed far better safety fitness records than the carrier population at-large, and tended to assign exceptionally experienced drivers to their vehicles, both LCV's and non-LCV's.
The findings of this study pertain only to the carrier population from which the sample was drawn. In this study, one cannot disregard the potential for self-reporting and selection biases.
The previous studies noted above have estimated a wide range of crash rates due to different databases, time frames, methodologies, and biases. Table VII-7 summarizes all the non-fatal and fatal crash rates from these various studies.
|Source||Time Period Analyzed||Type of Crash: |
|Type of Crash: Fatal|
|Longer Dimensioned Vehicle Study - FHWA||1983 - 1991|
|Analysis of Accident Rates of Heavy-Duty Vehicles - |
Campbell et al
|1980 - 1984|
|Tractor plus single trailer|
|Rural - Limited||4.50|
|Rural - Other||18.77|
|Urban - Limited||5.80|
|Urban - Other||14.32|
|Tractor plus double trailers (includes STAA Doubles)|
|Rural - Limited||4.06|
|Rural - Other||23.72|
|Urban - Limited||4.30|
|Urban - Other||13.98|
|Truck Weight Limits Issues and Options - TRB||1980-1984, presented for 1985|
|Tractor plus single trailer||245||10.20|
|Tractor plus double or Triple trailers (includes STAA Doubles)||269||11.20|
|Accident Rates of Multi-Unit Combination Vehicles Derived from Large-Scale Databases - Mingo et al||1986|
|Tractor plus single trailer||6.02|
|Tractor plus double or Triple trailers (includes STAA Doubles)||9.96|
|Long Combination Vehicle Safety Performance in Alberta - Woodrooffe||1995 - 1998||Collisions|
|Tractor plus single trailer||128.10|
|Rocky Mountain Doubles||16.55|
|Tractor plus double trailers||22.09||1.42|
|Tractor plus three trailers||107.35|
The reader should exercise care when comparing across studies since different data sources and definitions of variables were used in each study.
With the exception of the Woodrooffe study in Canada, the cited studies all rely on data that is many years old. The present study updates and focuses on the fatal involvement rates in the Scenario States by examining 1995 - 1999 fatal involvement and travel data. The data for number-of-crashes and number-of-trucks-involved came from the 1995 - 1999 Fatal Analysis Reporting System (FARS) final report. FARS provides data on the number of trailers for the combination vehicles involved in the crash and highway classification for all fatal crashes 28. Where a crash involved an unknown truck configuration or highway functional class, the crash was proportioned among the population of known crashes. The fatal crash numbers exclude single unit trucks and trucks not hauling a trailer (i.e. bobtails).
The 13 State VMT estimate is from the Highway Statistics VM-2 Table that lists the VMT by State and highway functional class. The splits between combination trucks and also between single-trailer and multi-trailer units utilize the detailed 1999 estimates of VMT by highway functional class prepared for this study and shown in Table II-4.
Although this represents more recent data than the previous studies, the analysis has many of the same limitations found in previous statistical safety analyses that attempt to estimate the respective safety of LCVs compared to other truck configurations. These include: (1) examination of past safety data may be an inaccurate predictor of future roadway safety; and (2) the analysis is unable to isolate LCVs from STAA doubles. Despite these shortcomings, the analysis demonstrates the importance of operating environment and potential trends.
Table VII-8 summarizes the fatal crash and travel data for 1995 - 1999 for the 13 States in the Western Uniformity Scenario. The data include 5 years of data to remove any bias that would be present in only examining a single year of data.
|Functional Class||VMT (million)||Fatal Crashes (number of Crashes)2||Fatal Crashes (count of Trucks involved)1|
|Single Trailer||Multi Trailer||Single Trailer||Multi Trailer||Single Trailer||Multi Trailer|
Sources: Fatality Analysis Reporting System (FARS), Highway Statistics VM-1 Table and 1999 expanded VMT prepared for this report.
1. Count of Trucks Involved contains all the trucks in a fatal crash. For example if two single-trailer trucks create a fatality then the entry for number of trucks involved is 2.
2. Number of Crashes contains the number of fatal crashes. For example if two single-trailer trucks create a fatality then the entry for the number of crashes is 1.
|Functional Class||Fatal Crash Rate (Number of Crashes)||Fatal Crash Rate (Number of Trucks Involved)|
|Single Trailer||Multi Trailer||Single Trailer||Multi Trailer|
*National crash rates were created using the same methodology and differences were found to not be significant at the 95% confidence interval.
Table VII-9 shows the fatal involvement rates given the VMT and fatal involvements in Table VII-8. Among the 13 States, the fatal crash involvement was 2.88 per 100 million VMT for single trailer combinations and 3.13 per 100 million VMT for multi-trailer combinations.29
Table VII-10 further develops the crash involvement rates from Table VII-9 by showing upper and lower bounds based on the 95 percent confidence intervals for single- and multi-trailer combinations on the different highway classes. A 95 percent confidence interval means that there is a 95 percent likelihood that the crash rate for a given year between 1995 and 1999 does not deviate from the mean crash rate for all years by more than approximately 2.0 times the standard error. For example, while the mean (or average) crash rate for multi-trailer combinations was 3.13, it could be expected - with 95 percent confidence - that the multi-trailer rate for a given year would fall between 2.42 and 3.84 crashes per million VMT. Similarly, while the mean crash rate for single-trailers was 2.88, it could be expected - again with 95 percent confidence - that the single-trailer crash rate for a given year would fall between 2.81 and 2.95 crashes per million VMT.
|Functional Class||Single Trailer||Multi Trailer|
|Lower Bound*||Fatal Crash Rate||Upper Bound*||Lower Bound*||Fatal Crash Rate||Upper Bound*|
*Lower and Upper Bounds are set by the 95% confidence interval.
It is tempting to look at Table VII-10 and conclude, among other things, that multi-trailer combinations are less safe than single trailer combinations. Tests for statistical significance show that such a conclusion would be incorrect.
This section discusses the data limitations that impede the prediction of fatal involvements under the Western Uniformity Scenario. Although quantitative estimates are not available, the Scenario may be judged in terms of the relative shifts that are projected to occur from: one configuration to another; the operating environments in which various types of LCVs would begin to operate; the relative stability and control characteristics of each configuration; the changes in truck travel miles that would result; the availability of qualified drivers; and the regulations that might be put in place to promote safe operations.
As noted above, the fatal crash and travel data do not allow a detailed examination of LCVs separately from multi-trailers. The multi-trailer classification largely contains data on twin 28-foot STAA doubles. According to an analysis of 1991-1996 data, LCVs comprise about 22 percent of the multi-trailer combination vehicles involved in fatal crashes,30 but there is no method to accurately estimate a separate fatal involvement rate. The measurement problem is three fold; fatalities are rare occurrences, there are few LCVs currently operating and there is only limited travel data collected on LCVs. There is no federal requirement to collect data for specific types of multi-trailer combination vehicles. Only 2 of the 13 Scenario States actively collect separate VMT for different types of multi-trailers.31
Without the ability to breakout the different multi-trailer types, the fatal involvement rates in Table VII-10 are too broad for predicting the Scenario's multi-trailer fatal involvement rates.
No attempt was made in the CTS&W Study to estimate changes in the number or cost of crashes that might result from any of the Scenarios analyzed in that study. Among the reasons why such estimates could not be made were (1) the weights and dimensions of many of the vehicles analyzed in that study were substantially greater than vehicles currently operating even in the West, (2) the LCVs were assumed to operate nationwide, including on highways with poorer roadway geometry and higher traffic volumes than on highways they currently use, and (3) uncertainties about the number of experienced drivers that might be available to operate LCVs considering the large increase in the number of LCVs.
In this Scenario, many of those analytical uncertainties are reduced. The Scenario vehicles are typical of vehicles already being operated in the Western States and the highway environment is the same or comparable to the environment in which LCVs currently are being operated. Despite that improvement one is unable to apply the multi-trailer fatal involvement rate to the estimated Scenario VMT since there is limited data on those LCVs currently operating. In addition there could be some uncertainty about the availability of drivers who are experienced in operating multi-trailer combinations,32 but not to the extent noted in the CTS&W Study.
Triples analysis is conspicuously absent from most prior studies and databases. Obtaining data on Triples travel is difficult since data is collected on tractors and the same tractor can pull either one, two or three trailers depending upon the shipper's needs. Only two of the reviewed studies included a separate analysis of Triples, the Alberta Study and Scientex's Accident Rates for Longer Combination Vehicles. The Alberta Study found Triples were involved in 107 non-fatal crashes per 100 million miles traveled. This is roughly 4.8 times the involvement of doubles. The Scientex Study calculated 829 Triple-trailer non-fatal crashes per 100 million VMT. These estimates are different by nearly an order of magnitude because their data was drawn from a low number of observations.
Triples currently operate in all the 13 analyzed States except Washington and Wyoming where Triples are not permitted. Technically, Nebraska does permit Triples, but in practice there are no Triples operating since they can only operate empty. In the Scenario States there were 11 Triples involved in fatal crashes for 1995 - 1999 but since triples are so infrequently involved in fatal crashes the number varies greatly from year-to-year.33 In 1995 there was only 1 triple-trailer combination involved in a fatal crash but in 1998 there were four.
The biggest challenge in triples fatal involvement analysis, similar to other multi-trailers, is estimating their travel. Since triples' VMT is so small relative to other truck configurations the exact numbers are difficult to derive from National or even State totals. As noted before, the Highway Performance Monitoring System (HPMS) that provides the best national data on truck travel does not include a classification for triples. Also, triples tend to be operated by less-than-truckload shippers who regularly drop and pick-up trailers from their terminals so on a given 1,000 mile operation one-half could be as a triple and one-half as a double. (This is different than the typical resource hauling LCV that remains as one multi-trailer unit for most operations.) Elsewhere in this study the VMT for triple-trailer combinations is estimated and utilized for impact analysis but due to the problems sited above one is unable to have confidence in an estimate of triple-trailer fatal involvement rate.
One is able to conclude, based on the stability and control properties discussed earlier in this chapter, that triple-trailer combinations have relatively poor dynamic stability in the present configuration. Woodrooffe (2001) suggests that Triple's performance could be improved "if coupled in the B-train or C-train configuration."
Safety is the primary factor when assessing potential changes in TS&W policy. Safety is the U. S. Department of Transportation's preeminent goal, State transportation agencies share this priority, and motorists who must share the road with large and heavy trucks would care strongly about the safety of those vehicles.
TS&W policy changes can affect safety in several ways. First, they can affect the total number of trucks on the road and thus the exposure of the overall truck fleet to crashes. Analyses of potential 2010 VMT changes for Scenario States indicate that the Scenario would reduce total heavy truck travel by between 9 and 25 percent. These figures include not only reductions in truck travel associated with shifts of freight from smaller to larger trucks that would be allowed in some States under this Scenario, but also increases in truck traffic caused by shifts of freight from railroads to trucks. Reductions in truck crashes would not be expected to be as large as reductions in travel for several reasons. First the greatest reductions in truck travel occur on the safest roads - rural Interstate highways. Since travel is not estimated to fall as much on other rural arterials that have much higher crash rates than rural Interstate Highways, the reduction in overall crashes would not be as great as the reduction in overall travel.
Most previous studies of potential safety impacts of changes in TS&W policy have relied primarily on studies that have compared crash rates of single- and multi-trailer combinations. As noted above the problem with this approach is that most multi-trailer combinations are short STAA doubles that are comparable in length and weight to single-trailer combinations. While these STAA doubles are less stable than standard single-trailer tractor-semitrailers when one looks at their rearward amplification and load transfer ratio, they perform better than tractor-semitrailers in terms of their static rollover threshold and offtracking. The various LCVs analyzed typically fall between the tractor-semitrailer and the STAA double in terms of stability and control properties. However, they are much longer and heavier than either of those standard vehicles and they have greater offtracking. These characteristics influence how easily a truck driver can maintain control should operating conditions become challenging or can regain control should it be lost in response to a precipitous event. These factors all make it difficult to extrapolate overall multi-trailer combination crash rates to the fleet of LCVs.
It is also difficult to extrapolate the results from studies conducted outside the U.S. because the operating environment may not be representative of the U.S. environment. Not only may highway and traffic characteristics be different than those in the U.S., but regulatory policies may also differ. Such regulatory differences could be expected to have a significant impact on the safety of LCV operations.
Even without the ability to estimate the potential changes in the crash rates that might be associated with operations under the Western Uniformity Scenario, it is useful to update the crash analysis for the 1995-1999 period, the latest years for which crash data were available. This update was needed since during the 1980's when most of the past studies were conducted the use of double-trailers was still growing and had not reached a steady-state equilibrium. Also vehicle safety in general has drastically improved since the 1980's with the advent of seat-belt requirements, air bags and anti-lock brakes34 - among many other things. From 1990 to 2000, the number of large trucks in fatal crashes per 100 million VMT declined from 3.3 to 2.4 - down 27 percent.35
The statistical analysis indicates the importance of operating environment. Among single trailer configurations the fatal involvements can range from 1.50 to 4.73 per 100 million VMT. Among multi-trailer configurations the fatal involvements can range from 1.39 to 6.36 per 100 million VMT. These numbers indicate that when estimating fatal involvements it is not just the magnitude of VMT change but on what road classes the VMT changes. In the Western Uniformity Scenario multi-trailer trucks operate 61 percent of their mileage on Interstate roads; if a larger portion of their mileage were to shift to non-interstate roads, one would expect the number of fatal crashes involving these vehicles to increase.
This chapter does not explore auto driver perceptions and reactions to LCVs.36 In surveys and focus groups conducted for the CTS&W Study, most drivers expressed concern about the safety of sharing the road with larger and heavier trucks. Any attempt to increase the size of trucks would require a major public education campaign on how to operate around large trucks and the relative safety enhancements that would be required of any new larger truck.
Campbell, K.L., et al. 1988. Analysis of Accident Rates of Heavy-Duty Vehicles, Final Report. University of Michigan Transportation Research Institute, Ann Arbor.
Fancher, P.S., Ervin, R.D., Winkler, C.B., Gillespie, T.D. 1986. A Factbook of the Mechanical Properties of the Components for Single-Unit and Articulated Heavy Trucks. Phase 1. Final Report. Michigan University, Ann Arbour, Transportation Research Institute. 190 p. Sponsor: National Highway Traffic Safety Administration, Washington, D.C. Report No. UMTRI-86-12/DOT/HS 807 125. UMTRI-74246
FHWA. 1987. An Interim Report on the Larger Dimensioned Vehicle Study. U.S. Department of Transportation.
FHWA. 1996. Accident Rates for Longer Combination Vehicles. U.S. Department of Transportation.
FHWA. 2000. Comprehensive Truck Size and Weight Study. U.S. Department of Transportation.
FHWA. 2003. Draft: Vehicle Performance Analysis. U.S. Department of Transportation.
GAO. 1992. Truck Safety: The Safety of Longer Combination Vehicles is Unknown. U.S. General Accounting Office.
Kostyniuk, Linda, et al. April 2002. Identifying Unsafe Driver Actions that Lead to Fatal Car-Truck Crashes. University of Michigan Transportation Research Institute sponsored by the American Automobile Association.
TRB. 1990. Special Report 225 Truck Weight Limits. National Research Council, Washington, D.C.
Mingo, R. D., et al. 1990. Safety of Multi-Unit Combination Vehicles. Association of American Railroads, Washington, D.C.
Roaduser International. 1999. Performance Based Standards for Heavy Vehicles in Australia. National Road Transport Commission.
Woodrooffe, J. 2001. Long Combination Vehicle Safety Performance in Alberta 1995 - 1998. Woodrooffe & Associates.
Woodrooffe, J. 2003. Vehicle Performance Analysis. Woodrooffe & Associates.