Regulatory Impact Analysis (4)

View a PDF of this document on Issuu by clicking here. Jump to the Table of Contents.

H.3.3.1 Variation among Models

Sleep models differ on how the various components interact with each other. Some models assume an additive interaction between the circadian and homeostatic components of alertness while later studies provide evidence of a nonadditive interaction.133 Jewett et al (1999)134 developed mathematical models in which levels of subjective alertness and cognitive throughput are predicted by three components that interact with one another in a nonlinear manner. These components are: (1) a homeostatic component (H) that falls in a s-shaped manner during wake and increases at a decreasing rate that asymptotically approaches a maximum during sleep; (2) a circadian component (C); and (3) a sleep inertia component (W) that increases at a decreasing rate after awakening.

Sleep models also vary depending on the model’s purpose. More recently, efforts have been made to develop sleep models that can quantify the relationship between sleep, circadian rhythm and performance. The models use sleep data as input and they yield predicted alertness as well as performance on monotonous tasks. Some models include an identification of levels at which the risk of performance or alertness impairment starts, as well as prediction of sleep latency and time of awakening of sleep episodes. Examples of this new generation of sleep models include the following: the Fatigue Audit InterDyne model (FAID), the Circadian Alertness Simulator (CAS) and the Walter Reed Sleep Performance Model (WRAIR-SPM).

H.3.3.3 Psychomotor Vigilance Task

The Psychomotor Vigilance Task is a test of behavioral alertness developed by David Dinges and John Powell in the mid-1980s at the University of Pennsylvania (UPENN) Hospital. The PVT was designed to evaluate the ability to sustain attention and respond in a timely manner to salient signals (Dinges & Powell, 1985).137 It was also designed to be free of a learning curve or influence from acquired skills, such as aptitude or education, and to be highly sensitive to an attentional process that is fundamental to normal behavioral alertness.

PVT performance has been demonstrated to be highly sensitive to detecting changes in behavioral alertness associated with numerous work settings, such as medical house staff jobs, night shift workers, drowsy drivers, transoceanic pilots. PVT performance is also sensitive to bodily states, such as those of partial and total sleep-deprived subjects, truck drivers with sleep apnea, sleepy elderly subjects, and to exposure to various ingested chemical substances, such as caffeine, modafinil, and alcohol.

In the study of operating practices in a high-fidelity truck simulator, O’Neill et al. (1999)138 examined two weeks of day-time driving (0700-2130 hrs) that entailed simulated driving tests of 12 hours per day on a 14 hours on duty and 10 hours off duty work schedule. Ten-minute PVT tests were administered three times per day (at 0645, 1330 and 2100 hrs) during a five-day driving workweek, and four times per day on the drivers’ weekend off recovery days (0900, 1300, 1700 and 2100 hrs). PVT data were reported in the form of median and mean reciprocal response times, and number of lapses, combined on a graph. The authors found that PVT scores were sensitive to partial and full sleep deprivation, thus underscoring the value of properly designed work-rest schedules.

The PVT is acknowledged as being one of the most consistently reliable research tools for the study of operator alertness, fatigue and/or drowsiness. The PVT test of simple choice reaction time is backed by almost two decades of experience and historical data. It has been used widely by the research community in many studies e.g. Balkin, (2000)139; O’Neill et al. (1999); Hartley et al., (2000)140 and Krueger, (2002).141

H.3.3.3.1 Model Calibration

Balkin et al. (2000) described the WRAIR-SPM as “a series of empirically derived mathematical relationships describing the continuous decrement of cognitive performance during wakefulness, restoration of cognitive performance during sleep, and cyclic variation in cognitive performance during the course of the day.”

Wakefulness was assumed to diminish cognitive performance capacity by a simple linear decay function Ct = Ct-1 – κw, where Ct is the cognitive performance capacity at time t, and κw is the performance depletion occurring in the interval t-1 to t.

Sleep was assumed to restore cognitive capacity utilizing an exponential growth function. For a subject going to sleep once cognitive capacity reached zero and remaining asleep for a period of time t, cognitive capacity would equal 100 * (1 – e-c2*t).142 In this representation, the coefficient c2 is the sleep recovery time constant.

The third component of the model is the circadian phase modulating function (M) which has both a circadian (24-hour) and ultradian (12-hour) component. To reflect the 24-hour circadian and 12-hour ultradian components, M is expressed as an additive double cosine function:

M = 1 + c3 * cos ((2π / 24) * t + c4) + c5 * cos ((2π / 12) * t + c6)143

where c3 and c5 represent the amplitude parameters for the cosine functions c4 and c6 represent phase shift parameters from midnight (the beginning of a day). In the WRAIR-SPM, predicted performance at a given time (t) is expressed as the product of the Current Cognitive Capacity (C) and the Modulating function (M).

H.3.3.3.3 Sleep/Wake History

The first major input to the model is sleep/wake history. The sleep/wake history represents the timing and duration of sleep and wakefulness periods over a period of days. In the WRAIR-SPM, four functions are used to relate sleep/wake history to cognitive performance capacity level (CPCL). These four functions, and a brief description of their relationship to cognitive performance are as follows:

Wake/Decrement Function – The wake/decrement function describes how cognitive performance declines during periods of continuous wakefulness.

Sleep/Restoration Function – The sleep/restoration function describes the rate at which cognitive performance capacity accrues during sleep.

Delay-of-Recuperation Function – The delay-of-recuperation function was incorporated into the model to exhibit the time lag between the wake/decrement function and the sleep/restoration function at the beginning of sleep. This delay is set at five minutes in the model, the time assumed to transition into recuperative sleep.

Sleep Inertia Function – The sleep inertia function accounts for the gradual restoration of normal performance and alertness upon awakening (approximately 20 minutes).

H.3.3.3.4 Time of Day (Circadian Phase)

The second input is the circadian phase, which is based on time of day. This component accounts for the empirical data showing that CPCL oscillates between a five and twenty percent peak to trough over a 24-hour period. Reflecting the influence of circadian and ultradian rhythms on performance, performance is lowest in early morning hours, and increases across the day (except for a dip in the afternoon), and peaks in the evening hours, prior to sleep onset.

Based on the user input of a sleep/wake schedule, the model will generate graphical and tabular outputs. The tabular data presents minute-by-minute reports of the subject’s sleep/wake status at the particular time and level of predicted performance. The level of predicted performance is reported numerically on a scale ranging from zero (0) to one hundred (100).

The model does not differentiate between “awake and working” versus “awake and resting” times. One might think that the former would take a greater toll on one’s performance level capacity for subsequent periods in the day. In addition, it does not explicitly take into account the interaction of physical and mental exhaustion and it does not recognize any time-on-task effects separate from the general cognitive depletion and circadian functions.

H.3.4 Estimating Sleep Profiles

The modeling uses actigraph data from the Walter Reed field study to predict sleep in a 24-hour period based on time on duty in 24 hours. The Walter Reed field study provides the most accurate measure of actual sleep (rather than reported sleep) as well as its relationship to time worked. Because the dataset follows a panel of people across time, the appropriate model is a random-effects cross-sectional time series model for panel datasets.144 Diagnostic tests indicate the appropriateness of a random effects model.145

The data suggest the most appropriate functional form is a cubic regression equation (third-order polynomial), particularly given the interest in accurately reflecting hours of sleep for those working longer hours.146 This relationship between time sleeping and time on-duty is then used to predict sleep given modeled numbers of hours on duty.147

Table 60: Predicted Hours of Sleep in 24-Hour Period by Selected Hours On Duty

Hours On Duty Predicted Hours Sleep
0 8.13
6 7.57
8 7.45
10 7.25
12 6.91
14 6.35
16 5.52
18 4.34
20 2.75

Source: Walter Reed Field Study.

Chart 5: Relationship of Sleep to Time Off


Source: Walter Reed Field Study

H.3.5 Estimating Change in SH short-haul: generally, < 150 mi. from base for property carriers Crashes using Model Schedules

Daily schedules were modeled for SH short-haul: generally, < 150 mi. from base for property carriers drivers to input into the Sleep Performance spreadsheet to predict differences in incremental crash risks for a baseline with full nights of sleep, actual compliance with current HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rules, current HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rules under full compliance, and the three proposed options. Research on schedules for SH short-haul: generally, < 150 mi. from base for property carriers drivers is more limited than for LH long-haul; generally >150 mi. from base for property carriers drivers. The literature indicates that these schedules are noted for their more regular pattern of work even if length of the work day varies

Daily schedules for SH short-haul: generally, < 150 mi. from base for property carriers drivers were modeled using information on typical SH short-haul: generally, < 150 mi. from base for property carriers schedules from the Virginia Tech Field Study and Virginia Tech Focus Groups. General lessons from these sources were confirmed with industry representatives involved in SH short-haul: generally, < 150 mi. from base for property carriers operations. Column 2 of Virginia Tech Focus Groups’ table 4 (p. 8) outlines typical daily patterns for local beverage truck drivers. The general pattern begins with an early start to the work-day beginning with pre-trip inspections and paperwork. The trip to the first delivery stop is about 15 to 30 miles followed by delivery and paperwork activities. This is followed by a series of shorter driving times (about three miles according to the table) among subsequent route stops and delivery and paperwork activities of roughly constant length. The route ends with a trip back to the facility followed by assisting in reloading or paperwork. The Virginia Tech Focus Groups’ table 14 (p. 77) lists average number of deliveries for beverage and snack delivery drivers as just over 11 per day. Figure 25 in the Virginia Tech Focus Groups (p. 75) indicates that the average SH short-haul: generally, < 150 mi. from base for property carriers driver in the focus groups spent 29 percent of their non-break time driving.148

All SH short-haul: generally, < 150 mi. from base for property carriers drivers are modeled as arriving at work at 7:00 am, the median response to time of day they start work. They are modeled as having awoken a half-hour before arriving at work. Sensitivity tests indicated that modeling the drivers as arriving at work 75 minutes after awakening made minimal differences. Following the information discussed above, SH short-haul: generally, < 150 mi. from base for property carriers drivers are modeled as performing non-driving work for their first half hour on duty, followed by a half hour of driving to arrive at their first delivery stop and a half-hour of non-driving work at the first stop.149 Drivers’ last half hour is modeled as non-driving work. For the remaining hours on-duty for SH short-haul: generally, < 150 mi. from base for property carriers drivers, time is modeled as a repeating pattern of alternating deliveries such that even numbered deliveries begins with a quarter-hour of driving followed by three-quarters of an hour of non-driving work and odd deliveries begins with a quarter-hour of driving followed by a half-hour of non-driving work. These lengths were chosen such that SH short-haul: generally, < 150 mi. from base for property carriers drivers drive just over 30 percent of the day for average workdays of 10.3 hours in length.150 Because FMCSA Federal Motor Carrier Safety Administration (The agency proposing the EOBR Electronic on-Board Recorder (A device attached to commercial motor vehicles that tracks the number of hours drivers spend on the road) rule) is interested primarily in the differences in crash risk among the proposals, the exact number of deliveries or length of time spent on each aspect of a driver’s duties is less important than the even distribution of driving throughout the day.

This general pattern of daily schedules is applied to 25-day schedules.151 The series of schedules are adapted such that the SH short-haul: generally, < 150 mi. from base for property carriers drivers do not work on the sixth and seventh day of the week to reflect the typical regular work week schedules found in the industry.152 For a given night, the amount of sleep is modeled based on the calculated relationship between sleep and time on-duty.153 Sleep was modeled for SH short-haul: generally, < 150 mi. from base for property carriers drivers as ending a half-hour before leaving for work. The time sleep begins varies according to the amount of total time (within a quarter hour) of estimated sleep. A 25-day working schedule was input from every fourth driver into the Sleep Performance spreadsheet for each of the proposal options as well as for the current compliance level. The schedules for each proposal vary only by the threshold at which individual work days are truncated.154 The resulting incremental crash incidence calculation for each scenario is subtracted from a baseline crash increment. This baseline increment represents the model’s estimated crash risk increment from five-day work week schedules of 8 hours of sleep with driving spread throughout the day.155 Given the assumptions in the baseline, the model provides a baseline crash increment of –1.72 percent, just slightly below the 0 percent expected theoretically. Because this baseline is subtracted from the increment for the 2003 rules, its size does not affect the results, which are presented at the end of the following section.

  1. LH Schedule Models
    1. Spreadsheet LH long-haul; generally >150 mi. from base for property carriers Crash Increment Calculations

After generating each rolling and non-rolling schedule modeled for each driver proportion cell, FMCSA Federal Motor Carrier Safety Administration (The agency proposing the EOBR Electronic on-Board Recorder (A device attached to commercial motor vehicles that tracks the number of hours drivers spend on the road) rule) calculated crash risk increments by entering the schedules into the Sleep/Performance spreadsheet. Results for driver schedules with stable working and sleeping patterns are displayed in table 61.

Table 61: Modeled LH long-haul; generally >150 mi. from base for property carriers Crash Increment Results, Stable Work/Sleep Pattern

Days Work / Week
Hours Work / Day 3 5 6 7 8
9 10% 14% 16% 18% 20%
11 13% 18% 20% 23% 26%
13 14% 21% 24% 28% 32%
15 23% 34% 41% 48% 57%

 

Source: RoutePro Simulations

  1. Weighting Crash Increments, Productivity and Proportion Fatigue-Related

The crash risk increments calculated above are multiplied by the percentage of drivers found in each cell in the driver schedule proportion matrices. The resulting value is subtracted from the baseline crash increment under schedules with eight hours of regular sleep for an interim crash increment score for the pre-2003 compliance status quo, pre-2003 rule with full compliance, and 2003 rule with full compliance.

These interim crash increments are adjusted for the differences in productivity found through these calculations from the productivity found in generating the cost estimates. Crash risk estimates are scaled up or down using the ratio of productivity found in the cost analysis to that found in the crash risk analysis. The results are multiplied by the proportion of truck crashes in which fatigue may have played a role. Only truck crashes in which truck driver fatigue is considered to have potentially played a role are included in this proportion. The productivity adjustments and the raw fatigue-related crash increments calculated across all cells in a driver proportion matrix are shown in table 62.

Table 62: Raw LH long-haul; generally >150 mi. from base for property carriers Crash Increment and Productivity Adjustment

Scenario Raw Crash Increment vs Baseline Productivity Adjustment Factor
Pre 2003 Status Quo 11.5% 0.0%
Pre 2003, Fully Enforced 8.4% 5.7%
2003 Rules 7.0% 0.0%

 

 

The final step discussed in this section is to convert the raw crash increment into the percentage of crashes that are related to fatigue, starting with the incremental crashes for each option that occur due to fatigue (raw fatigue increment). This number does not represent fatigue-related crashes as a proportion of all crashes, which ultimately is the proportion of interest. To calculate this proportion, termed the fatigue-related percentage, the raw fatigue increment is divided by the sum of the fatigue increment and the baseline of crashes before the fatigue increment. The baseline percentage of crashes before adding the fatigue increment is simply 100 percent. The sum of the raw fatigue increment plus the baseline percentage of total crashes is the raw fatigue increment plus 100 percent. The percentage of fatigue-related crashes, therefore, is the raw crash increments divided by 100 percent plus the raw crash increment or (raw increment)/(100 percent + raw increment).156

For the pre 2003 status quo scenario, the raw crash increment of 11.5 percent is divided by 100 percent plus 11.5 percent or 11.5 percent /(100 percent +11.5 percent) = 10.3 percent. The result of these calculations are shown in table 63, which also presents the equivalent fatigue-related crash results from the analysis of SH short-haul: generally, < 150 mi. from base for property carriers operations described above.

Table 63: Crash Increment and Fatigue-Related Crashes

Scenario Pre-2003, Full Compliance 2003 Rules Pre-2003 Status Quo
LH Raw Crash Increment vs.Non-Fatigued Baseline 8.4% 7.0% 11.5%
Fatigue-Related Crashes 7.8% 6.5% 10.3%
SH Raw Crash Increment vs.Non-Fatigued Baseline 3.6% 3.5% 3.7%
Fatigue-Related Crashes 3.4% 3.4% 3.6%

 

 

  1. Calibration of SP Results to Empirical Fatigue Crash Estimates

Because the SP spreadsheet is based on predictions of changes in simulated crashes rather than real-world experience, it cannot be used directly to estimate the percentage of crashes attributable to fatigue. Instead, that percentage was estimated independently. To ensure that they map well to the real world, the spreadsheet results need to be adjusted so that the scenario representing the status quo corresponds to this independent estimate of fatigue-related crashes.

The SP spreadsheet projected the fatigue-related crash percentage (relative to what would be expected for non-fatigued drivers) of 10.3 percent for LH long-haul; generally >150 mi. from base for property carriers operations and 3.6 percent for SH short-haul: generally, < 150 mi. from base for property carriers operations. The percentage of fatigue-related crashes is projected to be just under three times as great for LH long-haul; generally >150 mi. from base for property carriers as for SH short-haul: generally, < 150 mi. from base for property carriers operations. This difference is not surprising, given that SH short-haul: generally, < 150 mi. from base for property carriers drivers are much more likely than LH long-haul; generally >150 mi. from base for property carriers drivers to work during the day, sleep at home at night, and are less likely to be pushed to work extremely long hours. Previous research supports this general conclusion. For trucks on trips of 500 miles or more, the relative risk is even higher, at 2.35.”157

LH operations account for 61.8 percent of fatal truck crashes, 55 percent of injury-only truck crashes, and 59 percent of property-damage only crashes. Weighting by the number of crashes in each category, it was found that LH long-haul; generally >150 mi. from base for property carriers operations accounted for 58.2 percent of all crashes, with SH short-haul: generally, < 150 mi. from base for property carriers operations accounting for the remaining 41.8 percent. Fatigue accounted for 8.15 percent of all fatal truck crashes. Combining these percentages with the SP spreadsheet results showing fatigue-related increments of 3.6 percent and 10.3 percent for SH short-haul: generally, < 150 mi. from base for property carriers and LH long-haul; generally >150 mi. from base for property carriers operations respectively indicates that all of the estimates can be reconciled if the SP spreadsheet estimates are multiplied by an appropriate factor. This factor was found by setting up the following equation for X, the factor by which the SP spreadsheet estimates are to be multiplied:

41.8% * 3.6% * X + 58.2% * 10.3% * X = 8.15%

Rearranging terms and solving, X = 8.15%/(41.8%*3.6%+58.2%*10.3%)

= 8.15%/7.45% = 1.0917.

Thus, the SP spreadsheet can be calibrated to yield the 8.15% overall fatigue-related crash risk if the SP spreadsheet estimates of the crash increments are multiplied by 1.0917, producing fatigue-related increments of 3.9 and 11.2 percent for SH short-haul: generally, < 150 mi. from base for property carriers and LH long-haul; generally >150 mi. from base for property carriers respectively. Calibrating the estimates of the percentage of fatigue-related crashes for all of the options by multiplying by 1.0917 results in the estimates presented in table 64.

Table 64: Percentage of Crashes Attributable to Fatigue

Pre-2003 Rules 2003 Rules
LH Uncalibrated 7.8 6.5
Calibrated 8.5 7.1
SH Uncalibrated 3.4 3.4
Calibrated 3.8 3.7

 

 

  1. Risks Associated with New Drivers

A secondary impact of the proposed HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) options would be to change the number of relatively inexperienced drivers that operate in the trucking industry. Since there is evidence in the literature linking experience with accident rates, any changes in the number of inexperienced drivers would correspondingly change the overall accident rates for all drivers under the HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) options considered.

Calculations for the changes in crash rates for new drivers were performed using data from the UMTIP driver survey and the discrete time proportional crash hazards model estimated for drivers based on that data.158 Using the regression coefficients for experience and its squared term from that model, and data on driving experience from Abrams, et al. (1997),159 a crash risk as a function of driving experience was estimated. Chart 6 shows that relationship.

Chart 6: Effect of Experience on Crash Risk


The function above indicates a 28 percent average reduction in crash risk for existing drivers over their lifetime of driving. Given the coefficients on years of experience and its squared term, FMCSA Federal Motor Carrier Safety Administration (The agency proposing the EOBR Electronic on-Board Recorder (A device attached to commercial motor vehicles that tracks the number of hours drivers spend on the road) rule) also estimated a 10-year weighted average change in accident rates for new drivers. The 10-year time horizon was chosen to be consistent with the time period used in this analysis for the costs and benefits calculations.

The weights used for this calculation are based on the distribution of experience levels for new drivers. According to conversations with industry analysts, approximately 85 percent of new drivers come in without any driving experience outside of their training and the remaining 15 percent is estimated to have an average 4 years of driving experience.

There is evidence that suggests that the high turnover rates, especially in the TL segment, have been driven by the nature of the hours of service,160 among other factors. Conversations with industry experts on driver retention suggest that the proposed new rules could have a positive impact on turnover to the extent that they make the work schedules in this profession similar to some of the other blue-collar occupations. Experts also feel that the industry does not have adequate human resource programs to retain drivers, leading to the highest turnover rates within the first 12 months of tenure. If HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) compliance could bring about any reductions in drivers leaving trucking, it could reduce the need for hiring new inexperienced drivers161 and change the composition of the new driver pool such that it is on average more experienced. Moreover, according to some industry experts, there is a growing tendency among trucking companies to only hire new drivers with some experience. This trend can also increase the average level of experience for the new drivers, as well as change the composition of drivers with or without experience in the analysis. Because data are not available on reduction in turnover because of the proposed new rules or fraction of the companies that only hire new drivers with some experience, FMCSA Federal Motor Carrier Safety Administration (The agency proposing the EOBR Electronic on-Board Recorder (A device attached to commercial motor vehicles that tracks the number of hours drivers spend on the road) rule) looked at a case in which only 50 percent of the new drivers come in with no experience and the rest with 4 years of experience. The Agency also looked at an extreme case where 99 percent of the new drivers have no experience. The results of this analysis are presented in table 65.

Table 65: Estimated Crash Risk Changes for HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) Rules

Percentage share of new drivers with Total change in crash risk under
2003 Rules Pre-2003 Rules
No Experience 4-year Experience LH SH LH SH
50% 50% -0.10% 0.06% -0.21% -0.02%
85% 15% -0.26% 0.14% -0.55% -0.05%
99% 1% -0.33% 0.18% -0.68% -0.06%

 

Analysis of UMTIP and Abrams, et al, data.

Note that negative crash risk percentages imply that in that market segment, there is actually a reduction in overall crash risk because of a decrease in labor demand (or increase in labor productivity) from the HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rules.

The main conclusion from the table given above is that although the Agency did not expect an increase in crash rates if there is a need to hire new drivers, the relative increases in their crash risk probabilities are not that alarming. The table also suggests that the increase in crash rates for new drivers is not very sensitive to the composition of experience levels for new drivers, in that the changes in crash risks from the top row of the table to the bottom are generally much smaller than one percentage point whereas the risk reductions provided by the options are on the order of several percent. This analysis used row 2, the 85 percent – 15 percent division, to calculate the changes in dollar benefits.

  1. Value of Crash Reductions

The total damages from all large truck crashes can be found by multiplying the total number of crashes by the average damage imposed per large truck crash. The average value of damages per crash shown in table 66, $75,637, is based on research for the Department of Transportation.162 Multiplied together, the total number of crashes and the value of damages per crash yield total annual damages of over $32 billion.

Table 66: Calculation of Total Value of Large Truck Crashes by Year

Average per Year
Fatal Crashes 4,568
Injury Crashes 92,000
Property Damage Only Crashes 329,250
Total Large Truck Crashes 425,818
Average Damages per Large Truck Crash $75,637
Total Damages from Large Truck Crashes (millions) $32,208

 

Source: “Costs of Large Truck- and Bus-Involved Crashes,” Zaloshnja et al., table 10.

This total value of damages can be divided between LH long-haul; generally >150 mi. from base for property carriers and SH short-haul: generally, < 150 mi. from base for property carriers operations using the breakdowns of crashes by length-of-haul and severity. Dividing the total number of crashes of each severity level into LH long-haul; generally >150 mi. from base for property carriers and SH short-haul: generally, < 150 mi. from base for property carriers yields the total number of crashes for each length of haul. Multiplying these totals by the average value per crash yields an approximate value of damages from all LH long-haul; generally >150 mi. from base for property carriers and SH short-haul: generally, < 150 mi. from base for property carriers crashes. These estimates are only approximate because the damages per crash differ by crash severity, and the breakdown of crashes by length of haul differs according to the severity of the crash.

The last line of table 67 shows the effect of excluding two groups of LH long-haul; generally >150 mi. from base for property carriers drivers from the calculation of benefits. It is not expected that drivers in these two relatively small groups – team drivers (for both private fleets and for-hire carriers) and the LH long-haul; generally >150 mi. from base for property carriers drivers in LTL carriers – to have their work schedules significantly affected by changes in the HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rules. The changes in fatigue-related crashes estimated to result from the options would not apply to these drivers or to the crashes that involve them. The damage estimates for LH long-haul; generally >150 mi. from base for property carriers crashes are reduced by 14.6 percent, which is the estimate of the percentage of LH long-haul; generally >150 mi. from base for property carriers VMT accounted for by these drivers.

Table 67: Division of Crashes and Crash Damages by Length of Haul

LH % SH % LH Crashes SH Crashes Total
Fatal Crashes 61.8 38.2 2,823 1,745 4,568
Injury Crashes 55.0 45.0 50,600 41,400 92,000
Property Damage Only Crashes 59.0 41.0 194,258 134,993 329,250
Total Large Truck Crashes 58.2 41.8 247,681 178,137 425,818
Average Damages per Large Truck Crash $75,637 $75,637 $75,637
Total Damages (millions) $18,734 $13,474 $32,208
Total Damages, Excluding Largely Unaffected LH long-haul; generally >150 mi. from base for property carriers Drivers (Team and LTL LH) $15,999 $13,474 $29,472

 

Table 68: Damages Attributable to Fatigue by Option

Pre-2003, Full Compliance 2003 Rules Status Quo
LH Percentage of Crashes Attributable to Fatigue 8.5% 7.1% 11.2%
Total Damages of Fatigue-related Crashes (millions) $1,361 $1,138 $1,791
SH Percentage of Crashes Attributable to Fatigue 3.8% 3.7% 3.9%
Total Damages of Fatigue-related Crashes (millions) $506 $492 $528

 

 

  1. Results of 2003 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) Rule Analysis

Table 69 shows the results for the estimates of the change in the number of drivers, the primary determinant of HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) compliance costs.

Table 69: Changes in Drivers Needed for Full Compliance with HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) Limits

Percentage Change Pre-2003 2003
LH 8.1% 4.2%
SH 0.7% 1.4%
Numbers LH 121,500 63,000
SH 10,800 21,300
Total 132,300 84,300

 

 

The direct costs relative to the status quo are shown in table 70. This table shows the costs of the current rules with full compliance in the fourth column from the right. Because there would be costs for compliance with the pre-2003 rules, the costs of the current rules are higher relative to the status quo than relative to the pre-2003 rule with full compliance.

Table 70: Annual Direct Cost Changes for Full Compliance ($millions)

Cost Category Pre-2003 2003
LH Driver Labor Cost 1,185 550
Other Costs 769 332
Total Costs 1,954 882
SH Driver Labor Cost 143 233
Other Costs 90 168
Total Costs 232 400
Total Costs, LH long-haul; generally >150 mi. from base for property carriers and SH 2,187 1,282

 

 

Table 71 shows the benefits and adjusted benefits of compliance with the current and 2003 rule relative to the status quo.

Table 71: Annual Value of Crashes Avoided under Full Compliance ($millions)

Pre-2003 2003
Benefits of Avoided LH long-haul; generally >150 mi. from base for property carriers Crashes 429 653
Benefits of Avoided SH short-haul: generally, < 150 mi. from base for property carriers Crashes 22 32
Total Benefits of Operational Changes 451 685

 

 

Table 72: Annual Net Benefits for Full Compliance ($millions)

Pre-2003 2003
Total Benefits 443 671
Total Costs 2,187 1,282
Net Benefits -1,744 -611

 

 

Table 73 shows the effects of different fatigue-related crash percentage assumptions on net benefits relative to the status quo.

Table 73: Sensitivity of Net Benefits to Baseline Fatigue-Related Crash Percentage ($millions)

Pre-2003 Rules 2003 Rules
Net Benefits, 5% Fatigue Crashes -1,918 -876
Net Benefits, 8.15% Fatigue Crashes -1,744 -611
Net Benefits, 15% Fatigue Crashes -1,365 -35

 

 

  1. Changes to Cost and Benefit Calculations from the 2005 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) RIA

In the 2005, HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rule, FMCSA Federal Motor Carrier Safety Administration (The agency proposing the EOBR Electronic on-Board Recorder (A device attached to commercial motor vehicles that tracks the number of hours drivers spend on the road) rule) changed the regulations to constrain the use of sleeper berths to ensure that each sleeper berth period is at least 8 hours, and is supplemented by a 2-hour break that may be outside the sleeper berth. At that time, the Agency also implemented an exemption from maintaining RODS Record of duty status (A logbook maintained by CMV drivers to track driving time (i.e., duty status) for each 24-hour period) for certain SH short-haul: generally, < 150 mi. from base for property carriers operations, which generates an ongoing paperwork savings. However, compliance costs and safety benefits to SH short-haul: generally, < 150 mi. from base for property carriers were unaffected.

  1. Work Patterns

Patterns of working by drivers in the different sectors of the trucking industry for the basis for the 2005 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rule analysis. In particular, the analysis focused on intensity of effort; this may be thought of as the degree to which drivers work close to the limits imposed by the HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rules. This can refer to hours worked (on-duty hours) in a week and in a day, hours driven in a day, days worked and days off in a week. These measures are important for analysis of both productivity and safety effects of rule changes.

  1. Data Sources

The measures of work patterns and intensity were based on several data sources. There were four sets of data on current experience (under the 2003 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rule): data provided by Schneider National on some aspects of its operations; data from the Owner Operator Independent Drivers Association (OOIDA) based on a survey of its members; a survey of private carriers carried out by Professor Stephen Burks of the University of Minnesota; and data collected by FMCSA Federal Motor Carrier Safety Administration (The agency proposing the EOBR Electronic on-Board Recorder (A device attached to commercial motor vehicles that tracks the number of hours drivers spend on the road) rule) (the “field survey”). The Schneider, OOIDA, and Burks data were gathered with the express purpose of obtaining information on use of three aspects of the new rule: the 11th hour, restarts, and split sleeper periods. Each of these sources is focused on a different sector of the industry.

In addition to the above data, information was used from nine private interviews with carriers, eight small TL firms and one small LTL firm.

  1. Schneider

Schneider’s data are for a large TL firm and cover approximately 16,000 drivers. They were taken from company records for August and October of 2004.

  1. OOIDA

OOIDA data are based on owner-operators Self-employed commercial truck drivers or small businesses that operate trucks for transporting goods over highways for their customers and a few company drivers for TL firms. OOIDA posted a survey form on its website asking drivers for information on use of the 2003 rule provisions in June 2004. The data used here are based on responses from 1,223 drivers.

  1. Burks

Professor Burks mailed a survey form to private carriers asking for information on their drivers’ use of the new-rule features in June 2004. He received usable responses from 29 firms covering 3,311 drivers.

  1. FMCSA Field Survey

The field-survey data largely represent company drivers with small TL companies. In terms of distribution of company size, this makes sense; the most TL companies are quite small. In the field survey, 86 percent of for-hire, TL/OTR companies have fewer than 25 tractors. Viewed in terms of truckload company size, the field survey is a representative sample, but these small companies account for a fairly small share of TL/OTR VMT, about 17 percent. LTL firms and private carriers are sparsely represented in the field survey.

These data, based on drivers’ log books, were obtained from companies in the course of compliance reviews or safety audits. Data cover 542 drivers with 269 firms in the period July 2004 to January 2005. For each driver, data for one month of operation were collected.

  1. Average Hours Per Day—On-Duty and Driving

Two basic measures of work are daily hours of driving and total work, the latter term including all on-duty time, both driving and other work. The field survey and the Schneider data provide information on driving time per tour; only the field survey provides data on on-duty hours per tour. The field survey provides some information on local drivers; the Schneider data do not distinguish between local and OTR operations.

A basic assumption in the calculation is that a day is equivalent to a tour of duty. While there are exceptions, most drivers work one shift in a day. A tour of duty comprises the time from the driver’s start of work to end of work, including driving, other on-duty, and off-duty time. Results are in table 74. That the numbers for driving hours for Schneider and OTR drivers from the field survey are so close enhances confidence in these numbers, even though the Schneider data include local service along with OTR operation.

Table 74: Daily driving and on-duty hours—averages

Field Survey Schneider
Driving 7.7 7.6
On-duty 9.2 N/A

 

 

  1. Average Hours and Days of Work Per Week

For OTR drivers, a typical measure of work is number of hours in eight days, which shows how close drivers work to the 70-hour limit for eight days. A more complete understanding of drivers’ work patterns, though, is revealed by examining data on days worked per week. Both the field survey and the Schneider data give hours worked in eight days, 62 hours for Schneider drivers, 59 hours for field-survey drivers.163 Some intermediate steps are required to convert these numbers to days per week. They are first divided them by 9.2, the field survey figure for on-duty hours per tour of duty, to obtain days worked per eight days and then make a further adjustment to obtain days worked per seven days. These results are presented in table 75.

Table 75: Average Weekly Hours and Days Worked

Field Survey Schneider
On-duty hours/8 days 59 62
Days worked per week 5.6 5.9

 

 

  1. Intensity of Effort

Examining only the average hours of driving and hours and days of work might lead one to conclude that all drivers work well within the limits imposed by the HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rules. Many drivers work and drive longer hours than the averages, and this analysis relies on the percentages of drivers that work close to the limits for estimating productivity and safety effects. The following show the distributions of daily driving and on-duty hours and on-duty hours in 8-day periods.

Table 76: Driving Hours Per Tour of Duty

Driving Hours Percentage of Tours
Schneider Field Survey OOIDA
11 10.7 16.2 28.0
10 15.5 16.1 N/A
9 16.4 11.2 N/A
<9 57.4 56.5 N/A

 

The OOIDA survey asked for frequency of use of the 11th hour but did not otherwise ask about driving hours. The figure presented here was calculated from the underlying survey data.

It is worth noting that the on-duty hours show a pattern relative to the 14-hour limit different from that of the driving hours relative to the 11-hour limit. Drivers are driving ten or more hours in more than 25.0 percent of their work days while reporting 13 or more on-duty hours for only 8.0 percent of days. The latter number suggests that drivers are generally taking two hours of break in a 14-hour tour or their normal work shifts are shorter than 14 hours. The Agency suspects that both are true. Inaccurate logging of on-duty hours could also be a factor.

Table 77: On-duty Hours Per Tour of Duty (Field Survey Only)

On-duty Hours Percentage of Tours
14 2.7
13 5.5
12 13.2
11 15.6
<11 63.0

 

 

Table 78: On-duty Hours in 8-day Periods

On-duty Hours Percentage of 8-day Periods
Schneider Field Survey
Company Leased All TL
>64 41.0 41.3 26.3
60-64 23.5 25.7 16.6
50-59 35.6 33.1 57.1

 

 

Tables 77 and 78 show that, while daily on-duty hours tend to “bunch” away from the limit, multi-day on-duty hours bunch close to the limit, closer, indeed, than is the case for driving hours. Tables 76 and 78 give information on differences in behavior between company drivers and owner-operators. While driving hours show a marked difference, the difference in multi-day hours is slight. Some of this could be accounted for by the fact that OOIDA data include some owner-operators Self-employed commercial truck drivers or small businesses that operate trucks for transporting goods over highways for their customers working on their own authority; those in the Schneider data are all leased.

Regarding differences in the average driving hours listed in table 79, it should be noted that there are few owner-operators Self-employed commercial truck drivers or small businesses that operate trucks for transporting goods over highways for their customers in the field-survey data; the higher percentage of 11th-hour use from the field survey, as compared with Schneider, suggests that smaller companies may push harder than larger ones, insofar as the driving limit is concerned. The OOIDA data on the 11th hour could be seen as part of such a pattern, especially if one thinks the own-authority owner-operators Self-employed commercial truck drivers or small businesses that operate trucks for transporting goods over highways for their customers are using the 11th hour heavily. However, the multi-day hours show the reverse pattern. For 65.0 percent of reported instances, Schneider’s drivers have over 59 hours; from the field survey, the comparable number is 43.0 percent. This might suggest that a big company does not schedule as close to the driving limits as a smaller company might but enjoys greater success in marketing and, thus, is able to keep its drivers moving more consistently. There could, of course, be other explanations.

One must be wary of reaching too far in drawing inferences from these data. To the extent that data from different sources show consistent patterns, one can use this information in analysis with some confidence. One pattern that comes through consistently is that the preponderance of OTR drivers and trucking firms are not operating at, or close to, the HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) limits. Approximately 25 to 30 percent of drivers are driving more than nine hours regularly and 25 to 40 percent of drivers are regularly working more than 64 hours in eight days. The industry experts consulted for this analysis agreed that this is an accurate general view of industry operations.

  1. Use of New Provisions of the 2003 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) Rule

The analysis examined the use of three aspects of the 2003 rule: restarts, the 11th hour, and the split sleeper-berth provision. The data come from the sources already mentioned: Schneider, OOIDA, Burks, and the FMCSA Federal Motor Carrier Safety Administration (The agency proposing the EOBR Electronic on-Board Recorder (A device attached to commercial motor vehicles that tracks the number of hours drivers spend on the road) rule) field survey.

  1. Restarts

All four of data sources reported on use of restarts. OOIDA reported that almost 90 percent of drivers used the restart at least some of the time.164 Burks reported that private carrier drivers used the restart on 61.0 percent of their runs.165 Neither OOIDA nor Burks, however, reported on length of restarts.

A driver using the restart provision may not be taking only the minimum 34 hours for the restart period. Schneider and the field survey both reported a high level of use of restarts and gave information on the length of restarts. In Schneider’s data, only 2.0 percent of restarts were only 34 hours. Depending on the reporting period, one-quarter to one-third of the restarts were 44 hours or fewer. Forty-three percent were 58 hours or fewer. Schneider showed a bi-modal distribution with peaks at 39 and 62 hours. Presumably, the former reflects cases in which the driver has taken one full day off, plus a few hours from the preceding and following days; the latter would reflect two full days off. The field survey shows that 33.0 percent of restarts were 44 hours or fewer. This comports well with the Schneider data. On this basis, one can say that at least one-third of restarts are short enough to bring a productivity gain. Using the alternative method of the moving eight-day period, drivers would usually have to stay off more than 44 hours before returning to work.

Anecdotal information on company attitudes towards restarts is that they like the provision and find some productivity gain even though drivers are staying off more than 34 hours. Managers seem hesitant to demand a return to work after 34 hours, except in unusual situations. It may, of course, be the case that taking only 34 hours off would not fit with the work schedule of many drivers, that is, there would not be anything for them to do at the 35th hour. For example, the 35th hour might come at 3:00 AM, and the company might have no use for the driver until 8:00 AM. When a TL driver comes off his restart, his first task is to pick up a new load; the hour at which the company needs his services will be set by the requirements of the shipper of that first load.

  1. Split Sleeper Berths

Data clearly indicate that most drivers never split, and those that do do so only occasionally. Schneider data for October 2004 show 97 percent of drivers never splitting and only 0.4 percent splitting “regularly.” Before the new rule, Schneider did not allow solo drivers to split at all and has only allowed them to split on an 8-and-2 basis under the new rule. The data from OOIDA and the field survey show many more drivers splitting occasionally but few splitting frequently.

Table 79: Incidence of Splitting

Splitting frequency Field Survey OOIDA
0 times per month 66% 55%
1-4 times per month 20% 20%
0-4 times per month (sum of above rows) 86% 75%
Average percent splitting per day 6% 13%

 

 

The Burks data show a higher percentage of frequent splitting, although they are not directly comparable with those from the field survey and OOIDA. They suggest that 52.0 percent of drivers split four or fewer times a month with the rest splitting more frequently. It is not clear why private drivers would split more frequently than others. There might be a higher percentage of teams in Burks’s data; evidence suggests that teams split more frequently than solo drivers.166

The data in the following table come from an Insurance Institute for Highway Safety (IIHS) survey of drivers at weigh stations in Pennsylvania and Oregon and from FMCSA’s Driver Fatigue, Alertness and Countermeasures Study (DFACS).

The IIHS and DFACS findings agree that teams split more than solo drivers. There is some anecdotal evidence that the incidence of splitting by teams is higher than that found by IIHS and DFACS. Several comments to the HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rule docket suggested higher percentages than these and also indicated that team splitting is generally balanced; that is, sleeper periods and driving stints are about equal at four to six hours each.167 The IIHS/DFACS findings for solo drivers sometimes splitting are lower than those from OOIDA and the field survey.

Table 80: Incidence of Splitting—Team and Solo

IIHS DFACS
Solo 24% 22%
Team 47% 52%

 

 

Data on splitting clearly show that splitting for most solo drivers occurs on an occasional and opportunistic basis. They do not build splitting into their operating routines. When they do take a split period in the sleeper, they go right back to the ten-hour rest at the next rest period. This does suggest that most drivers find the limited rest period unsatisfactory and use it only to avoid some other problem. An unexpected period of congestion would be one example. However, routine splitting is probably part of the daily operation of many teams.

  1. The 11th Hour

Table 81: Percentage of Work Days with 11th Hour Use

Schneider OOIDA Field Survey Burks
10.4 28.0 16.2 31.0

 

Field-survey numbers for compliant drivers only. Burks data might overstate 11th hour use because drivers reported 11th hour use for “runs,” which may encompass an entire trip spanning multiple work days.

Data show that the 11th hour is definitely being used. Comparing Schneider data to OOIDA and the field survey implies that big companies use it less often than small companies or owner-operators. The analysis uses the assumption that usage is heavier for smaller firms and sets 25 tractors as the demarcation point between big and small TL companies. FMCSA Federal Motor Carrier Safety Administration (The agency proposing the EOBR Electronic on-Board Recorder (A device attached to commercial motor vehicles that tracks the number of hours drivers spend on the road) rule) estimates that 40 percent of TL VMT is from small companies and 60 percent is from large companies.

  1. Carrier Operation Simulations

In the simulation model used to assess impacts on the more complex types of carrier operations, a truck’s progress is tracked in a computer program as the driver moves between origin and destination points, choosing new loads at the end of each run from a set of choices randomly selected from a data base representative of inter-county shipment patterns. The driver’s choices are made on the basis of which loads feasibly can be picked up and delivered within specified windows, given the limits imposed by the need to stop and rest. Within feasible choices, the driver is assumed to choose (or be assigned) the load that is most advantageous in terms of its contribution to its productivity. Because the HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rules affect which loads can be delivered, and change the amount of time that can be devoted to driving, the model is able to estimate impacts on productivity, and the accompanying changes in typical schedules.

Operation Cycle of HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) Model

Your browser may not support display of this image.

The model starts at the user defined home terminal. Out of 20 randomly generated origin-destination pairs, it chooses the pair that best fits its schedule as well as maximizes its productivity. Then it moves to the origin terminal, waits until the pick-up window time, and loads the shipment. It then drives to the destination terminal, waits until the delivery window time, and unloads the shipment. At this point, the model again analyzes another set of 20 origin-destination pairs and repeats the same procedure prescribed above for the time duration defined by the user. The movement of the truck in the model is constrained by HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rules (i.e., all required rest periods) allowing the user to compare different facets of HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rules with assumption of full compliance.

At the end of the simulation, the model yields an output that shows how the CMV Commercial Motor Vechicles operator behaved at each time of day. The truck’s movement following the operation cycle is recorded in the schedule output table which reveals what the truck was doing at each time of the day each day during the whole simulation duration. The duration of simulation is defined by the user so the model can generate up to one year’s worth of the schedule table. Table 82 is a snapshot of the schedule table showing only a small portion of it. The table actually has over 40 columns providing details such as time of the day, day of the week, driving status, load status, origin county, destination county, cumulative duty, driving, and rest hours.

Table 82: Schedule Output of HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) Model


  1. LH Safety Benefits

Safety impacts were measured by feeding the working and driving schedules from the carrier simulation model into a fatigue model to project driver effectiveness levels, and then estimating the resulting changes in crash risks for different cases. Changes in fatigue-related crash risks, calibrated to match realistic levels, were then multiplied by the value of all affected crashes to yield estimates of total benefits.

The approach to this analysis is illustrated in the flow diagram below. The crash and benefit analyses use the output of the truck operations simulations as the starting point for the analysis. The operations analyses provide a series of realistic truck driver schedules for each trucking industry segment. The schedules specify driver activity for each half hour (off duty, on-duty driving, and on-duty performing other activities such as loading, unloading, and waiting) over a multi-day period. The outputs of the simulations are also used as inputs into cost modeling.

The simulation model does not provide estimates on how the driver splits off-duty time between sleep and other personal activities. A separate analysis to address this question was carried out to add this information to the working schedules, based on sleep pattern surveys and similar research. These analyses led to a set of algorithms for sleep time based on the length of the break and the time of day at the start and end of the break.

Flow Diagram for Crash Risk Reduction and Benefit Calculations


Your browser may not support display of this image.

The FAST/SAFTE human performance model, developed in part from research led by the Walter Reed Army Institute of Research, was used for the analysis. The model applies a large body of sleep and fatigue research, including circadian rhythms to provide an operator effectiveness percentage relative to a fully rested individual. The FAST/SAFTE model does not take into account TOT effects, so a separate analysis of these effects was performed to determine the relationship between TOT and crash risk. However, because the rule adopted in 2005 made no changes in the maximum number of driver hours per day, the results of the TOT analysis did not figure into the benefits calculations of the 2005 rule, but were used to evaluate alternatives that reduced maximum drive time to 10 hours.

  1. Off-duty Sleep Time Assignment

In order to use the FAST/SAFTE model to process the outputs of the operational model, it was necessary to determine how much sleep the drivers were getting and when that sleep would occur during a given off-duty period. The productivity analysis outlined above focused on the lengths of drivers’ on-duty, off-duty, and driving periods. While the safety model requires the length of the on-duty and off-duty periods, it also requires the amount of sleep taken by the driver, and the placement of that sleep within the off-duty period. These are the two functions of the sleep allocation model. After a driver’s schedule has been separated into on-duty periods, off-duty periods and sleep periods, it is ready for input into the FAST/SAFTE model.

The first step in the sleep allocation process is to determine how much sleep a driver is expected to get based on past work history. This calculation is a decreasing function based on the cumulative amount of on-duty time in the previous 24 hour period. The basic function is identical to the one used in the 2003 RIA. For a driver who works 14 hours a day on a continuous basis, that amounts to 6.57 hours of sleep per 24 hour period. Once the amount of sleep is determined, the model checks to see how much sleep the driver has received over the previous 24 hour period. If the driver has had more sleep than he is expected to get, a sleep surplus is assumed to exist. If the driver requires more sleep than he has received over the last 24 hours, he has a sleep deficit and the model allocates sleep until the driver’s deficit has been reduced to zero or until the driver begins his next on-duty period, whichever comes first.

The second step in this process is the actual placement of the sleep within the off-duty period. To begin, the model consolidates all of the driver’s sleep within a period of time. For off-duty periods less than 24 hours, it is assumed that the driver will rest in a single session, and so the sleep is consolidated into a single sleep period. For rest periods equal to or longer than 34 hours, the driver is assumed to be taking a week-end break or restart of some length, and multiple sleep periods will be allocated based on the length of the rest period. Once the sleep has been consolidated, it needs to be placed within the off-duty period. After some test runs involving different rest period lengths and times of day, it was assumed that the driver’s sleep period should be placed as late in his off-duty period as possible, while still allowing him to wake up 30 minutes prior to the beginning of his next on-duty period. This 30-minute buffer was included to allow the driver to overcome any sleep inertia present when he awoke. It was determined that by placing the driver’s sleep towards the end of his off-duty period, it allowed the start of the on-duty period with the highest possible level of effectiveness. Whether drivers base their personal sleep allocation decisions on this same rationality is not clear at this time.

  1. Importance of Regularity in Driver Schedules

Another observation from the results of the safety modeling was the importance of maintaining a ‘regular’ schedule, referring to the driver’s ability to work and rest in the same general timeframe over consecutive work days. The importance of regularity stems from the effect that circadian rhythm has on driver effectiveness. Those drivers that had substantial shifts in their daily work/rest cycle performed considerably worse than those drivers that maintained a relatively constant schedule. It should also be noted that those drivers that shift to an entirely new schedule and maintain it over a period of weeks will eventually adapt to the new circadian rhythm. It is those drivers that shift to a different schedule on a daily or weekly basis that show substantial drops in effectiveness. Chart 7 compares effectiveness (y axis) for regular versus variable schedules across the work week (x axis) as modeled in FAST/SAFTE.

Chart 7: Regular (left) Versus Variable (right) Schedule

The ‘regular’ driver, in addition to showing a higher overall effectiveness, also shows much less variability in effectiveness. The large drops in effectiveness shown in the output of the variable-schedule driver are a characteristic of a constantly changing schedule. In the two examples above, the average driver effectiveness over a one-year period for the ‘regular’ schedule driver was 92.95%. This compares very favorably to an average effectiveness of 77.89% for the driver with the variable schedule.

  1. Driver Effectiveness – Split v. Continuous Sleeper Berth Periods

Another important observation from the FAST/SAFTE model was the difference in driver effectiveness values based on how drivers took their off-duty periods, and specifically their sleep periods. Of particular interest were drivers who split their sleep period as compared to those that chose not to split. To model these two different drivers, the Agency used the FAST/SAFTE model to calculate the effectiveness of drivers with 10 hours of on-duty time and 14 hours of off-duty time each day. One driver was given the 14 hour off-duty period in one single block and the other driver was given two 7 hour off-duty blocks. Twelve simulations were run for each driver, each offset by 2 hours, to determine the combined effect of splitting and circadian rhythms. Four weeks of driver data were modeled for this particular analysis. In general, drivers who split their sleep period into two, 7-hour blocks had lower levels of effectiveness than those drivers that took one continuous 14-hour break. Chart 8 shows two screen shots from the FAST/SAFTE model comparing driver effectiveness for a continuous and a split off-duty period.

Chart 8: Continuous (left) Versus Split (right) Off Duty Periods

 

At different start times over the course of a 24-hour period, the driver that chooses not to split generally has a higher average effectiveness than the driver that splits. However, modeling shows that drivers beginning their shift between the hours of 22:00 and 0:00 show higher levels of effectiveness if they choose to split their rest period. Chart 9 illustrates these findings.

Chart 9: Driver Effectiveness at Different Start Times – With and Without Splitting


  1. Estimate of Raw Crash Increment

The primary source of data to form a link between crash risk and PVT scores produced by the FAST/SAFTE tool was a laboratory study carried out by Walter Reed Army Institute of Research, in which driving performance on a truck simulator was compared with PVT measurements for different levels of sleep deprivation. A robust straight line relationship between the log of crashes during a 45 minute driving session and fatigue level as measured by 100-PVT score was obtained. Chart 10 shows the scatter plot and the linear relationship. PVT scores were scaled so that a score of 100 indicates a fully rested individual.

Chart 10: Relationship of PVT to Relative Crash Risks


  1. Estimate Actual Fatigue-Related Crash Risk.

The final stem in the analysis is to convert the raw crash risk increments to an estimate of crash risk. This is achieved by calibrating the crash risk increments for a base case to real-world fatigue related crash data. The procedure is identical to that described used in the 2003 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) RIA. The raw crash risk increments are the percentage increase in crash risk over the crash risk for a fully rested driver. Thus the proportional change in fatigue-related crashes is represented by the ratio:

[100+crash increment for 2003 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) Rules]/[100+crash increment for 2005 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) Rules]

The base case for this analysis is the fatigue-related crash risk for LH long-haul; generally >150 mi. from base for property carriers truck operations under the 2003 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) regulations, estimated at 7% of all truck involved crashes. The fatigue-related crash risk percentage for each of the HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) scenarios analyzed in this analysis is then as shown below:

7.0 x [100+crash increment for 2005 Rule]

[100+crash risk increment for 2003 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) Rule]

  1. Crash Risk Results

The results of the crash risk modeling were found to be decrease in crash risk of 0.3 percent for LH long-haul; generally >150 mi. from base for property carriers operations. When these results are weighted by VMT such that experiences the real drivers can be extrapolated from the simulated drivers, the crash risk reduction was found to be 0.1 percent. This percentage was valued by multiplying it by an estimate of the total annual damage associated with LH long-haul; generally >150 mi. from base for property carriers truck crashes. For consistency with the 2003 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) analysis, the Agency used the value from the previous analysis of $32.2 billion in year 2000 dollars, or about $34.9 billion in year 2004 dollars. The 2003 RIA Regulatory Impact Analysis also presented an estimate of the percentage of total damages that were caused by the LH long-haul; generally >150 mi. from base for property carriers segment. Applying the same percentage – just over 58 percent – to $34.9 billion yields just over $20 billion. The reduction in risk attributable to the 2005 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rule, given this total value for all LH long-haul; generally >150 mi. from base for property carriers truck crash damages, is 0.1% * $20 billion or about $20 million per year.

  1. LH Compliance Costs

The analysis of costs recognizes that the different provisions of the options will affect carrier operations in complex and interacting ways. It also recognizes that these effects will depend strongly on the carriers’ baseline operating patterns, which vary widely across this diverse industry. To produce a realistic measurement of the options’ impacts, then, FMCSA Federal Motor Carrier Safety Administration (The agency proposing the EOBR Electronic on-Board Recorder (A device attached to commercial motor vehicles that tracks the number of hours drivers spend on the road) rule) divided the industry into broad segments, collected information on operations within these segments, and then created a model of carrier operations as they are affected by HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rules. The variety of operational patterns made it necessary to limit the analysis to the most important cases.

The model was first loaded with data representative of shipping patterns and carrier cost structures, and tested to ensure that it could realistically simulate typical lengths of haul, empty mile ratios, and productivity. It was then set up to cover most important cases, under constraints representing the options, and used to simulate carrier operations under different conditions. The Agency then analyzed the data representing the simulated operations, using changes in miles driven as a measure of productivity impacts. Output measures from individual runs were weighted to give a realistic representation of the affected industry, including the drivers’ use of the most important provisions. The weighted changes in productivity from this procedure were then used to estimate the cost increases imposed on the industry by the options, using an analysis of the changes in wages and other costs likely to result from changes in productivity. These productivity-related costs were combined with transition costs associated with shifting to new rules to produce estimates of total social costs.

For representative carriers in each of several carrier size categories, the financial impact of the HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rule was estimated in terms of the change in net income (in 2004$) to the carrier, as well as a change in profits as a fraction of operating revenues. The approach used to estimate these impacts involved the development of a pro forma financial model of firms of different sizes confronted by changes in productivity, wages, and prices. Financial impacts were estimated under two assumptions about prices of trucking services: unchanged prices (representing the short run), and prices after industry-wide cost changes have been passed through to consumers.

The 2005 rule resulted in small adverse financial impacts (reduced profits) on most carriers, directly related to the magnitude of the drop in labor productivity. The results in terms of profit impacts relative to revenues seem to suggest very small impacts for firms across the wide range of size categories examined, including both large and small entities. The threshold for impacts considered to be of moderate size is generally taken to be one percent of revenues, and the average impacts of the rule fall well below that magnitude.

  1. Core Cost Components

A significant portion of the total cost is driver labor costs. Changes in the number of hours drivers can work or drive were translated to changes in driver’s labor productivities using the simulation model explained above. These changes were then used to calculate the additional number of drivers needed to achieve full compliance. Changes in the number of drivers were then translated into labor cost changes using the estimated wage-hours worked functional relationship for truck drivers used in the 2003 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) RIA, described in section H.2.1.1.1 and table 49 above. The changes made in the 2005 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rules were also evaluated with respect to the same non-wage costs considered in the 2003 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) RIA, discussed in section H.2.2 above.

The results of the simulation model described in section H.5.3 above showed a 3.9 percent decrease in productivity for LH long-haul; generally >150 mi. from base for property carriers drivers (a 7.0 percent decrease for regional less a 3.1 percent productivity increase for LH long-haul; generally >150 mi. from base for property carriers OTR) using split sleeper berths as a result of the elimination of the sleep sleeper berth provision in the 2005 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rule. These percentages are calculated for the simulated drivers; to extrapolate to the real driver population, these figures were weighted according to the fraction of total VMT attributable to these drivers, resulting in a productivity loss of 0.042 percent (a 0.08 percent decrease for regional less a 0.038 percent productivity increase for LH long-haul; generally >150 mi. from base for property carriers OTR).

The 2003 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rule analysis found that a 3.9 percent increase in LH long-haul; generally >150 mi. from base for property carriers labor productivity from the 2003 rules could be valued at about $1 billion, or about $275 million per percentage point (referred to as the “unit cost”) in year 2000 dollars. The 2005 rule analysis updated this figure to $298 million per percentage point of productivity year 2004 dollars using the GDP deflator. Converting the total cost changes to a unit cost number, as is done here, is possible because the 2003 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) analysis showed that there was a linear relationship between changes in driver labor productivity and the associated costs. Table 83 presents the breakdown of LH long-haul; generally >150 mi. from base for property carriers unit costs.

Table 83: LH long-haul; generally >150 mi. from base for property carriers Unit A single vehicle in a motor carrier fleet. This does not include trailers Costs for HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) Rule Changes

Unit: Change in Labor Demand 1%
Change in Number of Drivers 15,000
Driver Labor Cost $176
Avoided Labor Wages -$429
Avoided Labor Benefits -$26
New Labor Wages $482
New Labor Benefits $149
Other Costs $121
Non-driver Labor $7
Trucks $50
Parking $15
Insurance $11
Maintenance $19
Recruitment $20
Total $298

 

 

  1. Training Costs for New HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) Rules

Several commenters provided data on costs of re-training drivers and other personnel on changes to the HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rules. Costs per driver varied between $75 and $150, and the Agency assumed $100 in its analysis of the 2005 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) rule. Using a 7-percent interest rate, 10 years as the amortization period, and 1.5 million total LH long-haul; generally >150 mi. from base for property carriers truck drivers (the same basis as for the 2003 RIA), it was calculated that the annualized re-training costs for the LH long-haul; generally >150 mi. from base for property carriers segment to be $21 million.

  1. Total Costs of the 2005 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) Rule Changes

Table 84: Total Costs of the 2005 HOS Hours of service (Regulations issued by FMCSA that limit the number of daily and weekly hours a CMV driver may drive) Rule

Change in LH long-haul; generally >150 mi. from base for property carriers Productivity 0.042%
Change in Annual Costs due to Productivity Impact=0.042*298 $13
Incremental Annualized Retraining Cost $21
Total Annual Incremental Cost $34