National Academies Press: OpenBook

Prioritization Procedure for Proposed Road–Rail Grade Separation Projects Along Specific Rail Corridors (2019)

Chapter: Appendix B - Methodology for Adjusting the USDOT Accident Prediction Value

« Previous: Appendix A - Literature Review by Category
Page 53
Suggested Citation:"Appendix B - Methodology for Adjusting the USDOT Accident Prediction Value." National Academies of Sciences, Engineering, and Medicine. 2019. Prioritization Procedure for Proposed Road–Rail Grade Separation Projects Along Specific Rail Corridors. Washington, DC: The National Academies Press. doi: 10.17226/25460.
×
Page 53
Page 54
Suggested Citation:"Appendix B - Methodology for Adjusting the USDOT Accident Prediction Value." National Academies of Sciences, Engineering, and Medicine. 2019. Prioritization Procedure for Proposed Road–Rail Grade Separation Projects Along Specific Rail Corridors. Washington, DC: The National Academies Press. doi: 10.17226/25460.
×
Page 54
Page 55
Suggested Citation:"Appendix B - Methodology for Adjusting the USDOT Accident Prediction Value." National Academies of Sciences, Engineering, and Medicine. 2019. Prioritization Procedure for Proposed Road–Rail Grade Separation Projects Along Specific Rail Corridors. Washington, DC: The National Academies Press. doi: 10.17226/25460.
×
Page 55
Page 56
Suggested Citation:"Appendix B - Methodology for Adjusting the USDOT Accident Prediction Value." National Academies of Sciences, Engineering, and Medicine. 2019. Prioritization Procedure for Proposed Road–Rail Grade Separation Projects Along Specific Rail Corridors. Washington, DC: The National Academies Press. doi: 10.17226/25460.
×
Page 56
Page 57
Suggested Citation:"Appendix B - Methodology for Adjusting the USDOT Accident Prediction Value." National Academies of Sciences, Engineering, and Medicine. 2019. Prioritization Procedure for Proposed Road–Rail Grade Separation Projects Along Specific Rail Corridors. Washington, DC: The National Academies Press. doi: 10.17226/25460.
×
Page 57
Page 58
Suggested Citation:"Appendix B - Methodology for Adjusting the USDOT Accident Prediction Value." National Academies of Sciences, Engineering, and Medicine. 2019. Prioritization Procedure for Proposed Road–Rail Grade Separation Projects Along Specific Rail Corridors. Washington, DC: The National Academies Press. doi: 10.17226/25460.
×
Page 58
Page 59
Suggested Citation:"Appendix B - Methodology for Adjusting the USDOT Accident Prediction Value." National Academies of Sciences, Engineering, and Medicine. 2019. Prioritization Procedure for Proposed Road–Rail Grade Separation Projects Along Specific Rail Corridors. Washington, DC: The National Academies Press. doi: 10.17226/25460.
×
Page 59

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.

53 Methodology for Adjusting the USDOT Accident Prediction Value This appendix describes the methodology used to iden- tify the variables to adjust the USDOT accident prediction value. The methodology is illustrated with the development data for crossings with gates when Scale Option A is used. A severity value of 5 for fatal accident, 3 for injury accident and 1 for property damage–only (PDO) accident was used in this scale. The variables were tested with a validation data set as well. A brief description of the development and validation data set is given in the next section, followed by a description of the methodology used in this study. Development Data The data set for the state of Illinois was downloaded from the FRA website and cleaned to remove missing or errone- ous values. Over the 5-year period, there were 370 accidents at 330 locations. Table B-1 gives the division based on the warn- ing device at each location. This data set was used for model development. Validation Data Data were also downloaded for four other states: Texas, South Carolina, Iowa, and Pennsylvania. The states were chosen so that the selected crossings would be distributed over the continental United States. The data were subjected to the same filters used for the development data. Over the 5-year period, there were 882 accidents at 735 locations. Table B-2 gives the number of crossings and accidents for the four- state data. States such as California and Washington were not chosen because the data-cleaning procedure filtered out the majority of the data points. Methodology Each of the site-related variables, individually as well as collectively, were used to adjust the accident prediction value. The variables that improved the ranking of the crossings A P P E N D I X B using the adjusted accident prediction value were identified. Table B-3 shows the top five crossings in Illinois when ranked by their severity value using Scale Option A. Of the gated crossings within the data set, 136 crossings have a severity value of 2 or more. The ranking of crossings using the USDOT accident prediction formula identifies 79 of the top 136 crossings. The 79 identified crossings have a cumulative severity value of 302. The proposed safety score should rank the crossings such that more crossings could be identified and the ones with a higher cumulative severity value identified. Adjustments to Gated Crossings Using Single Variable The six variables identified were used independently to apply adjustments to the USDOT accident prediction value. For each of the crossings in the database, an adjustment was made on the basis of the value of the variable and a weight, as shown in the equation below. adjusted USDOT accident prediction value for crossing USDOT accident prediction value normalized value of variablew = + p where w is the weight. For the variable crossing angle alone, there were 1,800 iter- ations (18 normalization schemes p 100 possible weights). The researchers wrote a Python script to expedite going through these trials. From these trials, those variables that showed improvements are given in Table B-4. The adjustment to the USDOT accident prediction value using variables distance to nearby highway intersection, maximum timetable train speed, and crossing surface could identify more crossings and gives a higher sum of the severity values. Tables B-5 to B-7 give the normalization schemes of the variable used for the adjustment and the corresponding weight.

54 Warning Device Number of Crossings Number of Crossings with Accidents Number of Accidents XB 1,964 55 58 FL 1,472 39 41 Gates 3,434 236 271 Table B-1. Number of crossings and accidents in development data (Illinois). Warning Device Number of Crossings Number of Crossings with Accidents Number of Accidents XB 3,574 140 161 FL 1,679 112 133 Gates 5,780 483 588 Table B-2. Number of crossings and accidents for validation data. Crossing ID Number of Accidents in 5 Years Number of Fatal Accidents Number of Injury Accidents Number of PDO Accidents Severity Value (F = 5, I = 3, P = 1) 386440H 4 1 1 2 10 608310D 3 1 1 1 9 289067H 2 1 1 0 8 079493L 2 1 1 0 8 372184D 3 1 0 2 7 Table B-3. Top 5 crossings in Illinois ranked by severity value. Variable Name Number of Crossings of Top 136 Crossings Identified by Severity Value (USDOT Formula Identified 79 Crossings) Sum of Severity Value of Identified Crossings (USDOT Formula Gives 302) Improved according to USDOT Accident Prediction Value? Crossing angle 78 299 No Distance to nearby highway intersection 82 311 Yes Number of tracks 79 300 No Maximum timetable train speed 82 311 Yes Highway speed 79 300 No Crossing surface 82 311 Yes Table B-4. Number and sum of severity values after applying adjustments (gated crossing). Distance to Nearby Highway Intersection Schemes <75 1 0 0 1 1 75–200 0.5 –0.25 –0.5 0.75 0.75 200–500 –0.5 –0.5 –0.75 0.5 –0.75 >500 –1 –1 –1 0 –1 Range of weight used for given scheme (w) 0.010 0.011 0.007 0.010 0.011 0.007 0.003 0.004 0.005 Table B-5. Normalization schemes and weights identified for variable distance to nearby highway intersection for model development data set (gated crossings).

55 To validate the above results, the same procedure was carried out on the validation data set that was created. In this data set, there are 224 crossings with a severity value of 2 or more. The ranking of crossings using the USDOT accident prediction formula identifies 128 of the 224 crossings. These 128 identified crossings sum up a cumulative severity value of 531. It was observed that crossings ranked using the adjust- ments due to distance to nearby highway intersection could identify 133 crossings with a cumulative sum of severity value of 557. The adjustments applied using maximum timetable train speed identified 135 crossings with a cumulative sum of severity value of 569 while the adjustments applied because of crossing surface identified 130 crossings with a cumula- tive sum of severity value of 547. The normalization schemes and the weights identified for the validation data set are given below in Tables B-8 to B-10. As seen from Tables B-5 through B-10, it was observed that adjustments applied using certain normalized values together with corresponding weights could show an improved ranking of the crossings on the basis of the severity of the accidents. Adjustments using maximum timetable train speed shows the most improvement, followed by distance to nearby high- way intersection and then crossing surface. The researchers decided to use combinations of the three variables identified to test for improvements. It was also decided to reduce the search space by limiting the weights to be tried by choosing a subset that contains all the w values between the highest and the lowest values identified for each variable from both the training and validation data sets. The reduction in search space was necessary; otherwise, the number of trials would be exponentially large. The combinations that were tried follow: 1. Maximum timetable train speed and distance to nearby highway intersection, 2. Maximum timetable train speed and crossing surface, and 3. All three variables. Maximum Timetable Train Speed ≤10 10–20 20–30 30–40 40–50 50–60 60–70 >70 Scheme 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Range of weight used for given scheme (w) 0.017, 0.018, 0.019, 0.020, 0.021, 0.022 Table B-6. Normalization schemes and weights identified for variable maximum timetable train speed for model development data set (gated crossings). Crossing Surface Scheme Unconsolidated 1 Timber 0.5 Asphalt 0 Rubber –0.5 Concrete –1 Range of weight used for given scheme (w) 0.007, 0.008 Table B-7. Normalization schemes and weights identified for variable crossing surface for model development data set (gated crossings). Maximum Timetable Train Speed Schemes <75 1 1 75–200 0.5 0.75 200–500 –0.5 –0.75 >500 –1 –1 Range of weight used for given scheme (w) 0.017, 0.021 0.005–0.013 Table B-8. Normalization schemes and weights identified for variable distance to nearby highway intersection for model validation data set (gated crossings). Maximum Timetable Train Speed Scheme ≤10 0.1 10–20 0.2 20–30 0.3 30–40 0.4 40–50 0.5 50–60 0.6 60–70 0.7 >70 0.8 Weight used for given scheme (w) 0.052 Table B-9. Normalization schemes and weights identified for variable maximum timetable train speed for model validation data set (gated crossings). Crossing Surface Scheme Unconsolidated 1 Timber 0.5 Asphalt 0 Rubber –0.5 Concrete –1 Range of weight used for given scheme (w) 0.016 Table B-10. Normalization schemes and weights identified for variable crossing surface for model validation data set.

56 Adjustments to Gated Crossings Using Multiple Variables For each of the crossings in the database, an adjustment was added on the basis of the combination of the variables considered, the normalized value, and the weight assigned. This adjustment was of the form adjusted USDOT accident prediction value for crossing USDOT accident prediction value normalized value of variable 1 w ii i i n p∑ = + = where wi is the weight corresponding to the ith variable and n is the number of variables used (in this case n is 2 or 3). Tables B-11 and B-12 below give the number of crossings identified and the sum of the severity values of the identified crossings for the model development and the model valida- tion databases. At this point, it was difficult to determine the appropriate adjustments that are to be applied because it was not possible to determine if the high severity crossings were ranked higher than others on the basis of the corrected USDOT accident prediction value. It is only known that more numbers of crossings were identified when the adjustment was applied than without it. Therefore, to visualize the ranking created by the various adjust- ments applied, plots were made to show the cumulative severity of the top ranked crossings with and without the adjustments applied. The researchers decided to look for the first quarter of the top crossings on the basis of the severity values to see if the top crossings were identified earlier in the ranking. The above two figures give the cumulative severity values plotted against the crossings sorted for the model develop- ment and model validation data set on the basis of 1. USDOT accident prediction value; 2. USDOT accident prediction value adjusted using maxi- mum timetable train speed; 3. USDOT accident prediction value adjusted using maxi- mum timetable train speed and crossing surface; Variable Name Number of Crossings Identified in Top 136 crossings per Severity Value (USDOT formula identified 79 Crossings) Sum of Severity Value of Identified Crossings (USDOT formula gives 302) Maximum timetable train speed and distance to nearby highway intersection 80 305 Maximum timetable train speed and crossing surface 78 308 Maximum timetable train speed, distance to nearby highway intersection, and crossing surface 84 323 Table B-11. Number of crossings identified and sum of severity values of crossings identified on model development database (gated crossings). Variable Name Number of Crossings Identified in Top 136 crossings per Severity Value (USDOT formula identified 128 Crossings) Sum of Severity Value of Identified Crossings (USDOT formula gives 531) Maximum timetable train speed and distance to nearby highway intersection 133 552 Maximum timetable train 134 558 speed and crossing surface Maximum timetable train speed, distance to nearby highway intersection, and crossing surface 127 535 Table B-12. Number of crossings identified and sum of severity values of crossings identified on model validation database (gated crossings).

57 C um ul at ive S ev er ity V al ue s of C ro ss in gs Position of Crossing in Sorted List variable Figure B-1. Cumulative severity values of crossings in model development data. C um ul at ive S ev er ity V al ue s of C ro ss in gs Position of Crossing in Sorted List variable Figure B-2. Cumulative severity values of crossings in model validation data. 4. USDOT accident prediction value adjusted using max- imum timetable train speed and distance to nearby highway intersection; 5. USDOT accident prediction value adjusted using maxi- mum timetable train speed, distance to nearby highway intersection, and crossing surface; and 6. Severity values (field data). From the above two plots, it is observed that the adjust- ments applied using maximum timetable train speed, dis- tance to nearby highway intersection, and crossing surface tend to identify the crossings with more severe accidents earlier in the ranking in both the model development and the validation data sets. The normalized value for the variables and the weights applied on these variables are given in Table B-13 below. Similar procedures were repeated for crossings with cross- bucks (XBs) and flashing lights (FLs). The procedure was also repeated with all severity scales listed in the report. The cor- rective variables identified for each of the cases are listed in the Tables B-14 and B-15 below.

58 Variable Name Categories Scheme Used Weight Applied Maximum timetable train speed ≤10 0.1 0.017 10–20 0.2 20–30 0.3 30–40 0.4 40–50 0.5 50–60 0.6 60–70 0.7 >70 0.8 Distance to nearby highway intersection <75 1 0.017 75–200 0.5 200–500 –0.5 >500 –1 Crossing surface Unconsolidated 1 0.011 Timber 0.5 Asphalt 0 Rubber –0.5 Concrete –1 Table B-13. Corrective variables normalized gated crossings (severity scale of 5, 3, 1). Variable Name Categories Scheme Used Weight Applied Maximum timetable train speed ≤10 0.1 0.047 10–20 0.2 20–30 0.3 30–40 0.4 40–50 0.5 50–60 0.6 60–70 0.7 >70 0.8 Posted highway speed ≤20 –2 0.005 20–30 –1 30–40 0 40–50 1 50–60 2 Crossing surface Unconsolidated 1 0.005 Timber 0.5 Asphalt 0 Rubber –0.5 Concrete –1 Table B-14. Corrective variables normalized for crossings with FL (severity scale of 5, 3, 1). Variable Name Categories Scheme Used Weight Applied Maximum timetable train speed ≤10 0.1 0.047 10–20 0.2 20–30 0.3 30–40 0.4 40–50 0.5 50–60 0.6 60–70 0.7 >70 0.8 Crossing angle <60 –0.25 0.005 ≥60 0.5 Crossing surface Unconsolidated 1 0.001 Timber 0.5 Asphalt 0 Rubber –0.5 Concrete –1 Table B-15. Corrective variables normalized for crossings with XB (severity scale of 5, 3, 1). Proposed Safety Score A safety score is computed for each crossing. The score has two components: safety score accident prediction value site-related adjustment 1 2 k k p ( ) ( )= + The first component of the safety module, accident predic- tion value, considers the predicted accident frequency for a location. The second component, site-related adjustments, is computed for each crossing on the basis of the characteristics of the crossing. Each of these components can be assigned a weight (k1 and k2) by the users to reflect the relative impor- tance of each component. Default values for k1 and k2 are 1. The equation for the safety score for XB, FL, and gated crossing when k1 = k2 = 1 are given below: Safety score for crossings with XBs when k1 = k2 = 1 safety score USDOT accident prediction value 0.047 normalized value for maximum timetable train speed + 0.005 normalized value for crossing angle 0.001 normalized value for crossing surface p p p = + + Safety score for crossings with FLs when k1 = k2 = 1 safety score USDOT accident prediction value 0.047 normalized value for maximum timetable train speed + 0.005 normalized value for posted highway speed limit 0.005 normalized value for crossing surface p p p = + + Safety score for crossings with gates when k1 = k2 = 1 safety score USDOT accident prediction value 0.017 normalized value for maximum timetable train speed + 0.017 normalized value for distance to nearby highway intersection 0.011 normalized value for crossing surface p p p = + + The normalizations and the weights used for Scale Option B with severity scales of 10 (for fatal), 4 (for injury) and 1 (for PDO) and for Scale Option C with 46.5 (for fatal), 1.7 (for injury) and 0.1 (for PDO) are given in the Tables B-16 to B-21 below.

59 Variable Name Categories Scheme Used Weight Applied Maximum timetable train speed ≤10 0.1 0.018 10–20 0.2 20–30 0.3 30–40 0.4 40–50 0.5 50–60 0.6 60–70 0.7 >70 0.8 Distance to nearby highway intersection <75 –1 0.001 75–200 –0.5 200–500 0.5 >500 1 Table B-16. Variables and weights for severity, Scale Option B (F/I/P = 10/4/1): gated crossings. Variable Name Categories Scheme Used Weight Applied Maximum timetable train speed ≤10 0.1 0.076 10–20 0.2 20–30 0.3 30–40 0.4 40–50 0.5 50–60 0.6 60–70 0.7 >70 0.8 Posted highway speed ≤20 –2 0.007 20–30 –1 30–40 0 40–50 1 50–60 2 Table B-17. Variables and weights for severity, Scale Option B (F/I/P = 10/4/1): crossings with FLs. Variable Name Categories Scheme Used Weight Applied Maximum timetable train speed ≤10 0.1 0.028 10–20 0.2 20–30 0.3 30–40 0.4 40–50 0.5 50–60 0.6 60–70 0.7 >70 0.8 Crossing angle < 60 –0.25 0.002 ≥60 0.5 Table B-18. Variables and weights for severity, Scale Option B (F/I/P = 10/4/1): crossings with XBs. Variable Name Categories Scheme Used Weight Applied Maximum timetable train speed ≤10 0.1 0.018 10–20 0.2 20–30 0.3 30–40 0.4 40–50 0.5 50–60 0.6 60–70 0.7 >70 0.8 Table B-19. Variables and weights for severity, Scale Option C (F/I/P = 46.5/1.7/0.1): gated crossings. Variable Name Categories Scheme Used Weight Applied Maximum timetable train speed ≤10 0.1 0.046 10–20 0.2 20–30 0.3 30–40 0.4 40–50 0.5 50–60 0.6 60–70 0.7 >70 0.8 Number of tracks 1 0.5 0.026 2 0.75 >2 1.0 Table B-20. Variables and weights for severity, Scale Option C (F/I/P = 46.5/1.7/0.1): crossings with FLs. Variable Name Categories Scheme Used Weight Applied Maximum timetable train speed ≤10 0.1 0.034 10–20 0.2 20–30 0.3 30–40 0.4 40–50 0.5 50–60 0.6 60–70 0.7 >70 0.8 Table B-21. Variables and weights for severity, Scale Option C (F/I/P = 46.5/1.7/0.1): crossings with XBs.

Next: Appendices C and D - The RCAT User Guide and Toolkit and Templates »
Prioritization Procedure for Proposed Road–Rail Grade Separation Projects Along Specific Rail Corridors Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s National Cooperative Highway Research Program (NCHRP) Research Report 901: Prioritization Procedure for Proposed Road–Rail Grade Separation Projects Along Specific Rail Corridors is designed to assist state and local planners in making prioritization and investment decisions for road–rail at-grade crossing separations.

The report provides a comprehensive means of comparing similar project alternatives within a specific rail corridor. Planning factors include economic, environmental, and community livability factors to support a robust decision process for making grade separation decisions.

NCHRP Report 901 also includes railroad crossing assessment tool (RCAT), a multicriteria evaluation tool that considers safety, economic, environmental, and community livability factors in a set of linked Microsoft Excel spreadsheets.

The report also includes a communications toolkit to help inform and convey to stakeholders and decision makers the relative objective merits of individual road–rail separation projects within corridors.

The assessment tool, communications toolkit, and user guide are published in electric only format as Appendix C - The RCAT User Guide, and Appendix D - The RCAT Toolkit and Templates.

During the past decade, railroad traffic has fluctuated in a number of key markets; coal traffic has declined, while other markets such as petroleum and intermodal have grown. Changing markets can impact the amount of rail traffic on rail mainlines, presenting challenges to state and local planners faced with making investment decisions about at-grade rail crossing improvements. This situation is particularly acute along urban rail corridors experiencing significant increases in train traffic or where the operating speed or train length has increased.

The traditional approach for making grade-crossing investment decisions has been guided primarily by the U.S. Department of Transportation, Federal Highway Administration Railroad–Highway Grade Crossing Handbook, which focuses heavily on traffic and safety factors. While safety continues to be a high priority in the development of road–rail grade separation projects, state and local decision makers need more robust criteria when competing against other projects for funding and construction.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!