Estimating the Safety Effects of Work Zone Characteristics and Countermeasures: A Guidebook (2018)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

14 Rationale Oftentimes, practitioners developing construction plans have options for accommodating traffic through the various phases or stages of the project. These alternatives can include factors such as whether or not to (a) close lanes or shoulders, (b) utilize narrower lanes on a tempo- rary basis, (c) close ramps, and (d) reduce acceleration or deceleration lane lengths. Similarly, decisions are sometimes made whether or not to incorporate safety countermeasures such as targeted law enforcement or work zone intelligent transportation system (ITS) technology into the TMP for a project. When making decisions, it would be useful to know the differences in expected crashes among these alternatives. Limitations Computing the relative difference in crashes expected for two or more work zone alternatives is fairly straightforward if (1) appropriate CMFs of the particular features, strategies, or counter- measures are available and (2) an appropriate baseline estimate of crashes is available to which the CMFs can be applied. Some CMFs predict the effect of an incremental change in the value of a particular feature (e.g., change in crash likelihood per 1-ft decrease in lane or shoulder width), while other CMFs simply predict the change in crashes if a feature is added or removed. However, work-zone-specific CMFs are often lacking for many features of interest. In such instances, all that can be done is to apply CMFs developed for permanent roadway features to what is expected, understanding that the results obtained are only a rough approximation of how an alternative may affect crashes during the time that it is in place in the work zone. Still, such approximations can often provide useful insight into the potential value of the different alternatives being considered. When comparing alternatives, it is important that practitioners remember to evaluate each one across the same time period. Often, one of the alternatives being considered will result in a work zone of shorter duration than the other. In these situations, it usually cannot be assumed that the alternative with the work zone of shorter duration will not experience any crashes dur- ing the saved time period relative to the other alternative. Rather, crashes will likely still occur at some normal (non-work-zone) frequency, as illustrated in the Estimating the Effect of Acceler- ated Construction on the Expected Number of Work Zone Crashes section in Chapter 2. Method The process of utilizing CMFs for trade-off analyses is fairly straightforward, as illustrated in Figure 8. The first step is to define the various work zone alternatives for which the expected C H A P T E R 3 Using CMFs to Evaluate Alternative Work Zone Design Features, Operating Strategies, and Safety Countermeasures

Using CMFs to Evaluate Alternative Work Zone Design Features, Operating Strategies, and Safety Countermeasures 15 safety effects are to be compared. Each feature, strategy, or countermeasure that differenti- ates between each alternative is defined. In addition, the expected length of the work zone or work zone segment affected by each alternative, expected duration of that alternative, and expected AADT are estimated. The second step is to determine the availability and suitability of CMFs, describing how the differences in features, strategies, and countermeasures affect crash expectancies. The applicability of the CMF for work zone analyses is also assessed at this time to help gauge the level of confidence expected in the results obtained. Those CMFs developed using work- zone-specific data are likely to be more applicable and trustworthy estimates than those developed for permanent situations. However, CMFs developed for permanent situations may be the best available at the time for evaluating the effect of a particular feature. A catalog of available CMFs that could be used for work zone safety analyses is provided in Chapter 4. Also included in that chapter are a number of features and potential countermeasures for which no CMFs currently exist. The available information regarding the known operational or other effects of those features is provided to assist practitioners who must make decisions on whether or not to utilize them. In the third step, baseline crash estimates are obtained or computed that will be used to evaluate each alternative. This is the baseline that the CMFs will be multiplied by to assess how the feature, strategy, or countermeasure will affect crashes. Some work zone CMFs have been developed relative to pre-work-zone conditions, whereas others were developed relative to a given set of baseline work zone conditions. It is important to determine what the baseline is for each CMF and to use that baseline estimate in the computations so that the best estimate of the incremental effect of the feature, strategy, or countermeasure can be computed. As part of this step, it is important to recognize that there may be instances where a CMF developed with non-work-zone data might still be applied to a during-work-zone baseline crash estimate. For instance, even though a lane width CMF was developed using non-work- zone data, an analyst might choose to apply it to a work zone baseline estimate. While the incremental effect of the CMF would be based on non-work-zone conditions, applying it to a baseline estimate of crashes expected to occur in general in the work zone would yield a greater absolute change in crashes than if it were applied to the normal non-work-zone crash estimate for that segment, because crash frequencies in segments with work zones tend to be higher than when no work zone is present. Limitations do exist as to the accuracy Step 1: Define work zone alternatives to be evaluated Step 2: Determine availability and suitability of CMFs for alternatives Step 3: Obtain appropriate baseline crash estimates for applying CMFs Step 4: Apply CMFs for each alternative to baseline crash estimate Step 5: Compute differences in crash estimates between alternatives Figure 8. Use of CMFs for assessing trade-offs of various work zone features, strategies, and countermeasures.

16 Estimating the Safety Effects of Work Zone Characteristics and Countermeasures: A Guidebook of available data and methods for estimating baseline work zone crashes on most facilities. Consequently, the accuracy of the results obtained using this methodology will only be as accurate as the baseline crash estimates used. The fourth and fifth steps of the procedure are multiplying the selected CMFs for each work zone alternative with its appropriate baseline crash estimate, and then computing crash estimate differences among the alternatives. The following section provides a few examples to illustrate this methodology. Examples of Computing Crash Estimate Differences for Alternative Work Zone Design Features, Operating Strategies, and Safety Countermeasures Estimating Benefits of an End-of-Queue Warning System An agency is contemplating the use of a work zone ITS to warn traffic about queues devel- oping during a 6-month bridge repair project on a roadway that crosses a four-lane Interstate facility. The contractor will perform nighttime lane closures (7 p.m. to 6 a.m.) on the Interstate to perform the work and will work 5 nights per week. The Interstate serves 70,000 vpd in this area, and queues are expected to develop each night of work. A traffic analysis indicates that traffic queues could grow to up to 5 miles on some nights. Normally, this section of Interstate experiences 14.8 crashes per mile per year, 50% of which occur during the hours when the work is scheduled to occur. The agency wishes to estimate how many crashes might be prevented if work zone ITS is incorporated into the project. The agency performs the following analysis. Step 1. Define work zone alternatives to be compared Alternative 1—perform the nighttime lane closures over the 6-month project without a work zone ITS. Alternative 2—install a work zone ITS at the beginning of the project to warn of queued traffic conditions when they occur. Step 2. Determine availability and suitability of CMFs for alternatives For both alternatives, a CMF will be needed to quantify the effect of the nighttime lane clo- sures on crashes. As given in Table 9, the CMF for working at night with one or more lanes closed is 1.61. For Alternative 2, a CMF describing the effect of the work zone ITS for queue warning is needed. From Table 16, the CMF for a work zone queue warning system installed when queues are expected is 0.56. Since the nighttime lane closures for this project are expected to result in queues, this CMF is appropriate for use in this analysis. Step 3. Obtain appropriate baseline crash estimate for applying CMFs The normal non-work-zone crash rate for this section of Interstate was determined to be 14.8 crashes per mile per year. The lane closures will be affecting 50% of these crashes over a 5-mile section of Interstate each night and will be occurring on 5 of the 7 nights each week. The queue warning CMF is measured relative to what would be expected to occur during the lane closure if the system were not used. Therefore, the baseline crash estimate for Alternative 2 would actually be the work zone crash estimate for Alternative 1. This is computed as follows: Baseline Crashes crashes mi yr mi mo mo yr days wk days wk night crashes all crashes crashes 14.8 5.0 6 12 5 7 0.5 1.61 21.28 ( ) ( )=         =

Using CMFs to Evaluate Alternative Work Zone Design Features, Operating Strategies, and Safety Countermeasures 17 For comparison purposes, the normal number of crashes expected to occur on this section of roadway on these nights if no work zone is present would be 21.28 crashes divided by the work zone CMF of 1.61, or 13.22 crashes. Thus, the nighttime lane closures are predicted to result in 8.06 additional crashes (i.e., 21.28 – 13.22 = 8.06) over the 6-month period of the project. Step 4. Apply CMFs for each alternative to baseline crash estimate As stated in the previous step, the CMF for nighttime lane closures was applied in the previous step. Consequently, the expected number of crashes under Alternative 1 is as follows: 21.281Crashes crashesAlt = For Alternative 2, the baseline crash estimate must be multiplied by the work zone queue warning CMF to compute the expected number of crashes: 21.28 0.56 11.922Crashes crashesAlt ( )( )= = Step 5. Compute differences in crash estimates between alternatives Subtracting the crashes expected under Alternative 2 from those computed for Alternative 1 yields the number of crashes that the installation of the work zone queue warning system is expected to avoid: Expected Crash Difference crashesAlt Alt 21.28 11.92 9.361 2 = − =− The use of a queue warning system at this location was computed to result in 9.36 fewer crashes. If desired, the agency could apply comprehensive crash cost numbers to estimate the societal benefit in comparison with the cost of the system. If, for example, the agency had found crash severities at past work zones to be distributed as shown in Table 2, the crash cost benefits of the queue warning system would be computed to be nearly $450,000. The crash cost savings estimate in this analysis assumes that the severity of crashes remains the same in either alternative and is equivalent to the overall average crash severity distribution that the agency had experienced in work zones. There is some evidence that the severity of crashes Crash Severity Level Proportional Distribution of Crash Severities Proportion of the 9.36 Crashes Reduced Average Crash Cost1 Crash Costs Saved in Alternative 2 Fatality (K) 0.005 0.0468 $4.509,991 $211,140 Disabling injury (A) 0.018 0.1685 $242,999 $40,940 Evident injury (B) 0.088 0.8237 $88,875 $73,205 Possible injury (C) 0.136 1.2730 $50,512 $64,300 Property damage only (PDO) 0.753 7.048 $8,325 $58,675 TOTAL 1.000 9.360 $448,260 1Crash Costs in the Highway Safety Manual, First Edition, updated to 2016 dollars. Table 2. Estimated crash cost savings if Alternative 2 is used in this example.

18 Estimating the Safety Effects of Work Zone Characteristics and Countermeasures: A Guidebook that occur when a queue warning system is implemented is also reduced relative to a no-system condition, which would further increase the crash cost savings that might be achieved (6). Comparing Narrowed Lanes to Temporary Widening An agency needs to do full-depth pavement repair in the right lane of a 0.5-mile section of a four-lane divided highway. The repairs create a pavement drop-off, so the agency plans to attach temporary concrete barriers to the rightmost 2 ft of the left lane through the project while repairs are made. The left lane is normally 12 ft wide. A 1-ft inside shoulder is normally present adjacent to the left lane. The agency is trying to decide whether to use the existing pavement width by reducing the lane width through the work zone to 11 ft and eliminating the inside shoulder, or to first add 2 ft of temporary pavement to the left edge of the shoulder so that a 12-ft lane and the 1-ft shoulder can be maintained during construction. The addition of the temporary pave- ment will add 2 weeks to the current 6-month project duration and an additional $300,000 to the project cost. The roadway normally accommodates 15,000 vpd, and the section of interest is estimated to experience 5.0 crashes per mile per year based on the agency’s calibrated SPF for that roadway. The agency performs the following analysis. Step 1. Define work zone alternatives to be evaluated Alternative 1—long-term right lane closure, reduction of left lane width from 12 ft to 11 ft, reduction of inside shoulder from 1 ft to 0 ft over 6 months. Alternative 2—2 weeks of daytime inside lane closures to construct temporary pavement (assume that approximately 33% of daily crashes occur during those work hours), followed by 6 months of the long-term right lane closure with the left lane width at 12 ft and a 1-ft inside shoulder. For the sake of this example, the initial 2-week period of Alternative 2 where the temporary pavement is being constructed will be denoted as “Part 1” and the 6-month pavement repair will be termed “Part 2.” Step 2. Determine availability and suitability of CMFs for alternatives Alternative 1—CMFs are needed for reducing a travel lane from 12 ft to 11 ft and for reducing the inside shoulder from 1 ft to 0 ft. In Chapter 4, lane width and shoulder width CMFs are listed in Table 32 and Table 17, respectively; however, these CMFs are of questionable applicability to work zones because they were developed for normal roadway segments. Note that for the shoulder width CMF, it is defined in the literature relative to adding width to the shoulder. Since the agency is considering reducing the shoulder width, the reciprocal of the CMF shown was used. The following CMFs were thus selected: =→ 1.0512 11CMFLane width ft = =→ 1 0.97 1.031 0CMFShoulder width ft In both alternatives, the right lane will be closed. The agency again examines the CMF catalog and determines that CMFs for daytime lane closures with workers present (in Table 8) are equal to 1.66, measured against the normal (non-work-zone) condition. For nighttime periods, the CMF for a lane closure with workers present is determined to be 1.61 (from Table 9). For Part 1 of Alternative 2, the CMF for daytime condition lane closures with workers present can be used, but only for the times during which the lane closures will be in place. The agency examines the time-of-day distribution of crashes that normally occur on that section of roadway and determines that 30% of crashes typically occur during the daytime lane closure hours planned.

Using CMFs to Evaluate Alternative Work Zone Design Features, Operating Strategies, and Safety Countermeasures 19 It is also assumed that work will occur Monday through Friday but not on weekends. Therefore, the lane closure CMF for Part 1 is computed as follows: WZCMF days days wk days days wk Lane closures Part 5 7 1.66 0.30 1.0 0.7 2 7 1.0 1.14, 1 ( )( ) ( )=    + +     = During Part 2 of Alternative 2, workers will not be present during all hours when the lane is closed. Given that the description of the daytime and nighttime lane closure CMFs indicate that workers were present, the agency decides that the 1.66 and 1.61 values are too high to apply during periods when work is not occurring. For comparison purposes, the agency decides to compute what the general work zone CMF (no lane closure) would be for this facility, using the CMF equation for four-lane facilities with worker presence unknown (shown in Table 7): 1.474 10.036 1.164 ln 15,000 11.231 1.248ln 15,000 WZCMF e e lanes = = ( ) ( )− − + − + Considering this value and the daytime and nighttime lane closure with workers present CMFs, the agency chooses to assume that a CMF of 1.60 for both daytime and nighttime hours throughout that portion of the project would be a reasonable value to use in the analysis. 1.60, 2WZCMFLane closures Part = Step 3. Obtain appropriate baseline crash estimates for applying CMFs To properly apply the CMFs determined in Step 2, separate baseline crash estimates are needed for Part 1 and Part 2. The value of Part 1 is as follows: 5.0 0.5 2 52 0.11Part1Baseline Crashes crashes mi yr mi wks wks yr crashes( )=     = For Part 2, the baseline crash estimate is as follows: Baseline Crashes crashes mi yr mi mo mo yr crashesPart 5.0 0.5 6 12 1.252 ( )=     = Step 4. Apply CMFs for each alternative to baseline crash estimate For Alternative 1, no CMF need be applied to the baseline crash estimate for Part 1. For Part 2, the baseline crash estimate is multiplied by the appropriate work zone CMF and the lane width and shoulder width CMFs, as shown below: 1.25 1.60 1.05 1.03 2.171, 2Crashes crashes crashesAlt Part ( )( )( )( )= = Therefore, the total estimate of crashes for Alternative 1 would be as follows: 2.17 0.11 2.281Crashes crashesAlt = + = For Alternative 2, the work zone CMF computed for Part 1 is multiplied by its baseline crash estimate, and the CMF computed for Part 2 is multiplied by its appropriate baseline crash estimate. Note that the lane width and shoulder width CMFs are not applied for this alternative. ( )( )= =0.11 1.141 0.122, 1Crashes crashes crashesAlt Part ( )( )= =1.25 1.60 2.002, 2Crashes crashes crashesAlt Part

20 Estimating the Safety Effects of Work Zone Characteristics and Countermeasures: A Guidebook Therefore, the total estimate of crashes for Alternative 2 would be as follows: 2.00 0.12 2.122Crashes crashesAlt = + = Step 5. Compute differences in crash estimates between alternatives The final step in the process is to subtract the crash estimates from each alternative to determine the expected differences in crashes: 2.28 2.12 0.161 2Expected Crash Difference crashesAlt Alt = − =− Thus, Alternative 1 is expected to result in an additional 0.16 crashes during the project, despite being of slightly shorter duration. If desired, the agency could compute a comprehensive economic cost of this estimate. For example, if the same distribution of crashes and crash costs as shown in Table 1 was assumed to be applicable here, the resulting crash cost savings would be computed as shown in Table 3. Based on the analysis shown, even though Alternative 2 has a slightly lower estimated crash frequency and associated crash cost savings, the amount of the savings is minor relative to the added $300,000 cost of providing the temporary pavement, and so would not be justifiable based on crash cost savings alone. Ideally, an analysis of the effect of the two alternatives on work zone capacity and traffic delays would also be performed (such as shown in Table 3), monetized, and added to this cost estimate. Crash Severity Level Proportional Distribution of Crash Severities Proportion of the 0.16 Crashes Reduced Average Crash Cost1 Crash Costs Saved in Alternative 2 Fatality (K) 0.005 0.00075 $4.509,991 $3,382 Disabling injury (A) 0.018 0.0027 $242,999 $656 Evident injury (B) 0.088 0.0132 $88,875 $1,173 Possible injury (C) 0.136 0.0204 $50,512 $1,030 Property damage only (PDO) 0.753 0.1205 $8,325 $1,003 TOTAL 1.000 0.160 $7,244 1 Crash Costs in the Highway Safety Manual, First Edition, updated to 2016 dollars. Table 3. Estimated crash cost savings if Alternative 2 is used in this example.