Roadside Hardware Replacement Analysis: User Guide (2021)

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.

16 THRC = ABCA(1 + IRR) → Step 22 = Step 20(1 + Step 22) The IRR is automatically calculated and shown in the box for Step 22. 3.6 INTERPRETING RESULTS The IRR calculated in Step 22 is the rate of return where the societal benefits accrued over the life of the replacement hardware exactly equals the cost of removing the existing hardware and installing the upgraded hardware. Larger IRR values indicate a higher rate of return and are, therefore, a better use of scare highway agency funds than projects with lower values of IRR. Recently, the United States gross domestic product (GDP) growth has been about two percent annually. The most optimistic estimates of GDP growth for the next several years are just under four percent annually. A highway improvement project should have an IRR greater than the actual GDP growth and probably should be above the optimistic GDP growth values. If the IRR is less than about four percent, the replacement project is probably not a good way to spend scare highway agency resources. The IRR can also be viewed as a measure of priority, projects with the highest IRR should have the highest priority since they generate the largest societal benefit for the funds expended. A replacement project with an IRR = 8 would be a much better use of highway agency funds than another project with an IRR = 6 for example. Project with negative IRR values would be a poor use of funds under any economic conditions. 4 EXAMPLE 4.1 REPLACING STRONG-POST W-BEAM GUARDRAIL The following example both illustrates the roadside hardware replacement analysis approach to analyzing the need for roadside hardware replacement while also addressing one of the most important particular cases: evaluating the need to replace 27-inch tall w-beam guardrail and its associated terminals with 31-inch tall w-beam guardrail. Assume a highway agency is planning a 6.2-mile-long mill and resurface project on a two- lane rural highway with an AADT of 1,450 vehicles/day. The project is what has often been described as a 3R project and involves no new construction or reconstruction. The roadway alignment will not be changed, the basic nature of the roadway will not change, and only roadside hardware that is already in place will be considered for replacement. The total cost of the project is estimated as $920,186 not including any hardware replacement. Aside from signs, the roadside hardware on the anticipated 6.2-mile long project consists of 1,775 ft of 27-inch tall w-beam guardrail with 18 Report 350 guardrail terminals. The guardrails and terminals are in generally good condition and an inspection of the highway segment indicates the guardrails and terminals are crash-ready and located properly according to local highway agency policies. In this situation, should the 27-inch w-beam guardrails and Report 350 terminals be replaced by 31-inch tall w-beam guardrails and MASH terminals? The following sections will examine this question for this particular roadway. The filled in workbook book for this example is shown in Figure 2.

17 4.1.1 Steps 1-8: Risk Assessment of Existing Guardrails First, the performance of the existing 27-inch w-beam guardrails must be established by examining crash records or an in-service evaluation. Based on 2010 through 2015 Pennsylvania Department of Transportation (PennDOT) police-reported crash data there were 1,711 passenger car crashes with guardrail face Type D installations (i.e., 27-inch tall strong-post w-beam with blockouts) and 689 pickup truck and sport utility vehicles (SUV) crashes with Type D guardrail face installations. The total number of passenger vehicle Type D guardrail face crashes, therefore, was 2,400 (i.e., 1,711+689=2,400). Of the 2,400 total Type D guardrail face crashes, 17 were fatal or serious crashes so the proportion of fatal and serious injury crashes for all types of passenger vehicles was 0.0071 [0.0050, 0.0100] (note: the values in brackets in this section are the bounds of the 85th percentile CI). For passenger cars, the proportion of fatal and serious Type D guardrail face crashes was 0.0053 [0.0033, 0.0085] and for pickup trucks and SUV the proportion was 0.0116 [0.0071, 0.0192]. The proportion of fatal and serious crashes was higher (almost twice) for pickup trucks and SUVs although the result was not statistically significant at the 85 percent CL (i.e., the confidence intervals overlap). According to the PennDOT roadway inventory, there were 6,683 miles (35,286,240 ft) of Type D guardrail in the state in 2016. According to the FHWA highway statistics for Pennsylvania, the average AADT on a PennDOT maintained roadway was 2,260 vehicles/day so there were 6,683‧365‧6‧2,260/106=33,077 MVMP guardrails during the same data collection period as the six years of crash data. The crash rate for Type D guardrails was therefore the 2400 Type D guardrail face crashes divided by the traffic volume passing guardrails or 2400/33,077=0.0726 Type D guardrail face crashes/MVMP. 4.1.2 Steps 9-11: Relative Risk Estimates for Replacement Guardrail Next, the relative risk of the proposed replacement 31-inch tall w-beam guardrails must be estimated. Since 31-inch guardrails are relatively new, there is little if any field data available to use in assessing the field performance. Since the field performance is not known, the engineer must use judgement and develop an estimate based on the best available information. In this case, the motivation for developing and deploying 31-inch tall w-beam guardrail is to attempt to reduce the number of penetration and rollover crashes associated with pickup trucks and SUVs. This seems reasonable based on the PennDOT data since pickup trucks and SUV have a fatal and serious injury crash proportion that is more than double that of passenger cars though it is still very small and it is not definitively known if the difference is due to rollovers and penetration. The assumption in upgrading from 27- to 31-inch tall guardrails is that the 31-inch tall guardrails will reduce the percent of fatal and serious injury pickup truck and SUV crashes to the same level as passenger cars (i.e., 0.0053). As shown in the last section, the fatal and serious injury proportion of Type D guardrail face crashes was 0.0053 [0.0033, 0.0085] for passenger cars, 0.0116 [0.0071, 0.0192] for pickup trucks and SUVs and 0.0071 [0.0050, 0.0100] for all passenger vehicle types. If 31-inch guardrail was successful in reducing the KA ROR crash rate with type D guardrail faces such that the performance of pickup trucks and SUVs was the same as passenger cars, the risk reduction would be (0.0071- 0.0053)/0.0071 = 0.2518 or a 25 percent relative risk reduction. It is important to recognize that this is an estimate of the expected performance based on the design objective of the crash tests and engineering judgement but only a review of crash records in the future can positively confirm that the relative risk reduction was real.

18 4.1.3 Steps 12-17: Project Details for Guardrails Now that the relative risks of the existing and potential replacement guardrails are known or estimated, the specific details of a project can be examined. As discussed earlier there are 1,775 ft of w-beam guardrail in the 6.2-mile long project. Recent bid prices indicate Type D guardrail costs between 15 and 20 $/ft so 17 $/ft for materials and installation was assumed. Similarly, removing the existing guardrail costs about 2 $/ft based on recent bid prices. The cost of removing the existing 1,775-ft of w-beam guardrail and replacing it with 31-inch guardrail is, therefore, $33,725. In the last section, the police-reported Type D guardrail crash rate was found to be 0.0726 crashes/MVMP and replacing Type D guardrail with 31-inch guardrail would result in a 25.18 percent relative risk reduction. If it is assumed that the frequency of crashes remains the same (i.e., no major changes in traffic mix or volume or roadway alignment) it is reasonable to expect 0.0726‧0.0071‧0.2518‧(1,450‧365‧1,775/5,280‧106) = 0.0000323 fewer fatal and serious injury w- beam guardrail crashes/yr on the project segment after 31-inch guardrails replace the existing 27-inch tall guardrails. 4.2 REPLACING REPORT 350 TERMINALS WITH MASH 2016 TERMINALS Guardrails like the Type D guardrails in Pennsylvania discussed in the previous sections, are not replaced in isolation. Every guardrail has at least one terminal and an anchor so if the guardrail height is raised the terminals, anchors and transitions will also need to be modified. This section examines the risks and costs associated with replacing the associated terminals on the 6.2-mile long example project. 4.2.1 Steps 1–8: Risk Assessment of Existing Terminals As for the guardrail, the first task in examining the effectiveness of replacing the guardrail terminals is assessing the existing guardrail terminal performance. Unfortunately, the inventory of terminals in Pennsylvania is not as extensive as the guardrail inventory so it could not be used to match police reports to terminal installations. Instead, terminal crash rate was estimated based on data from another state, Ohio. The Ohio DOT 2009-2011 guardrail and guardrail terminal inventory indicates that there were 19,724 Type E guardrail terminals on state-maintained roadways (19). Type E guardrail terminals at the time were all tangent Report 350 terminals. Ray reported that in 2002 through 2012 there were 286 Type E terminal crashes, eight of which were fatal or serious injury crashes so the proportion of Type E terminal crashes that were fatal or serious was 0.0280 [0.0158, 0.0490]. The crash rate for terminals also needs to be estimated. Ideally, the number of vehicles passing terminals should be known for the same data collection period and area as the crash data. The best way to calculate this would be based on an inventory linked to traffic data as described earlier for Step 8. Unfortunately, such a linkage is not available for the Ohio data so it must be estimated instead. In 2009 there were 113,673 MVMT on 122,926 miles of state-maintained roads so the average AADT of a roadway in Ohio was 2,533 vehicles/day (20). The estimated exposure for terminals would be approximately the average AADT multiplied by 365 days/year multiplied by the 19,724 terminals on public roads times the 11-year data collection period or 365‧2,533‧19,724‧11/106=200,592 MVP Type E terminals. The crash rate was, therefore, approximately 286/200,592 = 0.0014Type E terminal crashes/MVP.

19 4.2.2 Steps 9–11: Relative Risk Assessment of Replacement Terminals Quantifying the improved performance in the field of MASH 2016 guardrail terminals is challenging. The changes to the test matrix of MASH from those in Report 350 with respect to terminals were relatively minor. In addition to both the small and large passenger vehicle increasing slightly in weight, the gating tests were changed from 15 degrees to a range of 5 to 15 degrees. Testers are instructed to choose the most critical impact angle. Many of the MASH terminals are slight reworkings of the prior Report 350 terminals. More to the point, while the FHWA and AASHTO found that all Report 350 energy absorbing terminals had three performance limitations (i.e., side impact, shallow-angle corner impact, and high-energy impact crashes), MASH does not address any of these in its test recommendations or evaluation criteria so MASH terminals may have similar performance limitations to the Report 350 terminals (21). For the sake of this example, assume that the MASH terminals result in a 50 percent reduction in the fatal and serious injury crash risk. Clearly, this is simply a guess that assumes MASH terminals are much more effective than Report 350 terminals. This is also an example of making an assumption that favors the replacement option. If the procedure results in a recommendation to not replace the hardware, not doing so is likely the correct use of funds since the replacement option was favored but still failed. If MASH terminals do in fact perform at this level, replacement hardware risk would be 0.0140. 4.2.3 Steps 12–17: Project Details for Terminals As discussed earlier, there are 18 guardrail terminals on the 6.2-mile long project. The cost of removing each existing guardrail terminal is estimated to be about $600 and the cost of installing a new MASH 2016 terminal is estimated as $2,000 so the replacement cost is 2,600 $/terminal. The total cost of replacing the 18 terminals would be $46,800. Assuming the replacement terminals result in a 50 percent risk reduction ratio and the terminal crash rate is 0.0140 terminal crashes/MVP as calculated in the last section, the replacement of the terminals should result in 0.0280‧0.50‧0.0014∙18‧1,450‧365/106 = 0.000190 fewer fatal and serious injury terminal crashes annually. 4.3 ECONOMIC ANALYSIS For this example, project, replacing 1,775-ft of 27-inch guardrail with 31-inch tall guardrail would cost $33,725 and replacing the 18 Report 350 terminals with MASH terminals would cost $46,800. The total cost of the replacement hardware would, therefore, be $80,525. According to the FHWA guidance on economic analysis, the 2020 VSL is $12.3 million (22). The annual benefit of replacing 27-inch guardrail with 31-inch guardrail and replacing Report 350 terminals with MASH terminals is the VSL multiplied by the number of fatal and serious crashes avoided. The previous sections estimated that 0.000023 fatal and serious guardrail crashes would be avoided annually, and 0.000190 fatal and serious terminal crashes would be avoided annual for a total of 0.000213 fatal and serious injury crashes are expected to be avoided annually by replacing the current hardware with MASH hardware. The annual benefit of the replacement project is, therefore, expected to be 0.000214‧12,300,00 = $2,626. In this example, the IRR is –1.51 percent over the 25-year design life. In other words, for the economic and site conditions of this example, replacing the existing hardware with MASH tested hardware is not a good investment of highway agency resources. In fact, the investment required to replace the existing hardware with MASH hardware would take more than 30 years to achieve an IRR of zero.

20 4.4 DISCUSSION The results of the analysis of the example problem discussed in the previous sections is sensitive to the assumptions that form the basis of estimating the replacement hardware performance. By definition, the new hardware likely has no field observed performance so estimating the performance will often be based on a combination of engineering judgement and intuition based on crash test results. The advantage to this method, however, is that it is possible to examine these assumptions by performing “what-if” analyses. For example, in the previous example the effectiveness of both 31-inch tall w-beam guardrail and MASH guardrail terminals had to be estimated. For the conditions used in the example, the IRR was negative indicating the benefit would never equal the cost of construction during the design life. In the workbook, the estimated hardware effectiveness is easily changed. For example, the MASH terminals would have to perform such that the proportion fatal and serious injury crashes was less than 0.0100 for the IRR to just equal zero. This would represent a more than 64 percent risk reduction ratio from Report 350 terminals to MASH terminals. Given the relatively modest changes in MASH compared to Report 350 for guardrail terminal crash tests and the similarities between Report 350 and MASH terminal technologies, a 64 percent risk ratio reduction seems unlikely. Similarly, 27-inch w-beam guardrail already performs at a very low risk (i.e., 0.0071) so reducing it significantly seems unlikely. Increasing the risk reduction ratio to over 90 percent still results an IRR of near zero. Little improvement can be expected in replacing still functional guardrail for this modest AADT two-lane rural roadway. On the other hand, using the original risk estimates presented in the example (i.e., terminals with a proportion of fatal and serious injuries of 0.0140 and guardrails with 0.0053), changing the traffic conditions can make replacement more attractive. For example, if the AADT is changed from 1,450 vehicles/day to 3,000 vehicles/day the IRR is 4.5 percent, which is essentially at the breakpoint of deciding to recommend the project. A highway agency could use an analysis like this, for example, to develop a policy to only replace existing functional crash-ready roadside hardware with replacement MASH hardware on rural two-lane roads with AADTs over 3,000 vehicles/day. The IRR increases to over 10 for AADT greater than 5,000 vehicles/day. Even though the analysis results in this example indicate replacing the existing hardware with MASH hardware is not a good use of funds for that particular roadway, a highway agency also has other considerations to balance. The particular roadway may have a crash history that is not well represented by the results or there may be public pressure to replace certain roadside hardware. Similarly, engineering judgement has an important role to play and an agencies experience with particular local design issues may overrule a purely economic decision. The roadside hardware replacement analysis presented here should not be considered the sole basis for a definitive decision but one consideration among many for determining when and where roadside hardware is best replaced and where it is best left in place.

21 Figure 2. Worksheet for Considering the Replacement of 27-inch W-Beam Guardrail with 31-inch W-Beam Guardrail and Associated Terminals. Intructions: Enter information on the project or policy being considered in the yellow shaded cells. Values in grey shaded cells are default values but can be changed if desired. Project Name: Risk Assessment of Existing Hardware Symbol 27" Type D GR R350 Type E 1. Enter the number of years of crash data: YCD 6 11 2. Enter the total number of crashes with the existing hardware: NCj 2400 286 3. NFSCj 17 8 4. Enter the desired cofidence interval (e.g., 85th percentile): CL 85% 85% 85% 5. REHJ 0.0050 < 0.0071 < 0.0100 0.0171 < 0.0280 < 0.0460 < < 6. Enter the quantity of existing hardware in crash reporting area: QEHJ 35,286,240 ft 19,724 ea 7. EEHJ 33,077 MVMP 200,592 MVP 8. Calculate the total police reported crash rate. CREHJ 0.0726 crashes/MVMP 0.0014 crashes/MVP Relative Risk Estimate for Replacement Hardware 31" MGS GR MASH 9. Estimate of risk of fatal or serious injury given a crash with the replacement hardware: RRHJ 0.0053 0.0140 10. Are the existing and replacement crash risk outside the 85th confidence interval? No Yes 11. Calculate relative risk reduction of replacement hardware compared to existing hardware: RRRJ 0.2518 0.5013 Project Details 12. Enter the design year AADT of the project: AADT 1,450 veh/day 1,450 veh/day 1,450 veh/day 13. Enter the amount of hardware installed on project: QEHJ 1,775 ft 18 ea 14. Enter the unit installed cost of the replacement hardware: UCIHJ 17.00 $/ft 2000.00 $/ea $/ea 15. Enter the unit cost to remove existing hardware: UCRHJ 2.00 $/ft 600.00 $/ea $/ea 16. Calculate the total cost of installing the replacement hardware: CRHJ 33,725 $ 46,800 $ - $ 17. Estimate the number of annual fatal and serious injury crashes avoided: FCSAJ 0.000023 crashes/yr 0.000190 crashes/yr crashes/yr Economic Analysis 18. Calculate the total cost of the hardware replacement: TCRH 80,525 $ 19. Enter the value of statistical life (VLS) or use the default value of $12.3 million: VSL 12,300,000 $ 20. Calculate the annual societal benefit of the replacement: ABCA 2,626 $/yr 21. Enter the assumed service life or retain the assumed 25 year life. YL 25 yrs Type CTRL+SHIFT+R 22. Calculate the internal rate of return assuming a 25 year design life. IIR -1.51 % to Clear and Reset Form ROADSIDE HARDWARE REPLACEMENT ASSESSMENT WORKSHEET The expected number of fatal and serious injury crashes avoided calculated in the previous step are summed and assigned an economic value based on the FHWA's value of statistical life (VSL) recommendation. The internal rate of return where the sum of all the benefits in future years equals the construction investment in the design years is calculated. Generally, if the internal rate of return is less than 2% it is not the best use of funds. Enter information about the proposed project. Use the project information along with the values determined in the previous sections to estimate the number of fatal and serious injury crashes avoided by replacing the existing hardware with upgraded hardware. Replace 27-inch PennDOT Type D guardrail and R350 Type E terminals with 31-inch MGS guardrail and MASH 2016 terminals Using state or district wide police reported crash data determine the number of fatal and serious injury crashes that occurred involving the existing hardware. If an in-service evaluation of the hardware is avaiable use the crash severity performance from that source. Use as many years of crash data as are available and enter the appropriate values in the yellow shaded cells. Three columns are provided to account for associated roadside hardware. For example, replacing a guardrail may also require replacing the associated terminals, anchors and transitions associated with that guardrail. Place information for each type of hardware in one of the three columns. Estimate the smallest likely risk of a fatal or serious injury with the replacement hardware. Generally, crash data or an in-service evaluation will not be available to determine this value so it must be estimated using engineering judgement. It should be less than the observed value for the existing hardware and must be greater than zero. Enter the estimates for each type of replacement hardware below: Enter the existing hardware exposure during the data collection period (i.e., million vehicle miles passing [MVMP] for guardrails and million vehicles passing for terminals [MVP]). Calculate the mean risk of fatal or serious injury with the existing hardware with 85th percentile confidence: Enter the number of fatal and serious injury crashes with the existing hardware: