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.
107Â Â As stated earlier in this report, each study was developed with detailed steps to help agencies follow the protocol. Five studies were not piloted. However, they include the necessary detail to enable any interested agency to follow the steps and implement the study. The following section discusses the studies that were not piloted. 10.1 Study 5: Asset-Level Risk IndexâNon-NBIS Culverts This study, though not piloted with a DOT, used data from the North Carolina DOT (NCDOT). It illustrates an example of the development of a culvert risk index for small (non-NBIS) cul- verts. The methodology was developed for implementation in a configurable PMS. The general methodology presented here follows the asset-level risk approach also used in Studies 4 and 6. The scenario and guidance developed for testing include the following: â¢ Threats: Flood events and hurricane impacts on culverts. â¢ Consequences: Roadway becomes impassable because of non-NBIS culvert failures (e.g., washouts or flooding). â¢ Implementation Effort: Define risk index input values for a subset of small (non-NBIS) cul- vert assets. Define how the risk index fits into objective function. Run analysis to optimize a subset of culverts, then analyze and summarize the resulting metric projections. 10.1.1 Study Objective and How the Results Help Inform Asset Risk Decisions The objective of this proof-of-concept study was to demonstrate a risk management method- ology for managing the vulnerability of non-NBIS culverts as well as risk to associated portions of roadway caused by washouts and/or flooding during storm and hurricane events. This proof of concept was developed for application in existing pavement or general asset management systems. Similar to the pilot study for pavement section flooding, this study was conducted at the asset level and illustrates the methodology for defining an asset-level risk index for use in a DOTâs PMS. Non-NBIS culverts for the purposes of this study generally include pipes and cul- verts greater than 48Â in. in diameter and below the threshold of a bridge (20Â ft measured along the centerline of the roadway, though multiple lines of smaller diameter could be considered). The assumption in this case was that roadway washouts or flooding would most often take place along streams, creeks, or drainage ways where small culverts are placed. 10.1.2 Description of the Technique or Tool This study used existing inventory and condition data collected for non-NBIS culverts to assess the vulnerability of these structures to flood events. This vulnerability rating could then S E C T I O N 1 0 Protocols Developed for Studies That Were Not Piloted
108 Risk Assessment Techniques for Transportation Asset Management: Conduct of Research be combined with identified threat probabilities and consequences to create a risk index for each structure. As with the pavement section risk study, the implementation of the process outlined in this study takes advantage of any existing vulnerability assessments that the agency has already conducted or is conducting. Once the index is developed, the study exhibits a proposed method for incorporating risk into an existing PMS by aggregating the risk index values into a section culvert risk index for all culvert locations on a pavement management section. Alternatively, the index could be incorpo- rated into a general asset management system in which each culvert location would represent an individual asset. Given the additional benefit in terms of risk, the study illustrates how projects that include risk mitigation could receive higher priority for selection based on the benefitâcost optimization process. 10.1.3 Methodology Used in Conducting the Study This proof-of-concept study follows the same eight steps (listed in TableÂ 10-1) as Studies 4 and 6. This section provides a summary of the process; more details are outlined in AppendixÂ H. The steps detailed in TableÂ 10-1 were followed with example data from NCDOT to simulate what an implementation of the methodology would entail. Input Data Used in the Study This proof-of-concept study was modeled based on data provided by NCDOT from a survey of non-NBIS pipes conducted across the state. The survey data were used to assess the vulner- ability of sections of roadway, including culvert locations, to various storm events and combined with traffic data to assess potential delay and repair consequences. This made it possible to evalu- ate overall risk at each location. Similar to the road section flooding study, this study used as input quantified values for each threat, for each culvert location, for the following: â¢ Threat probability â¢ Vulnerability of the culvert location to washouts or flooding given the threat occurs â¢ Consequence of washout or flooding Methodology Steps Step 1 Identify available sources within the organization of vulnerability studies that have been conducted for the identiï¿½ied threat or hazard type. Identify whether quantiï¿½ied threat probabilities, vulnerabilities, and consequences are available from an existing assessment. Deï¿½ine values for these for each vulnerable location within the study scope. Step 2 Determine and conï¿½igure in the management system the calculation of the quantified risk index component as a function of the threats, vulnerabilities, and consequences. Step 3 Decide on any deterioration models for changes in the threat probability, vulnerability, and consequences over time, possibly as the result of climate change. (Note that assumed deterioration rates can be constant.) Step 4 Identify risk mitigation actions (projects) and the associated reduction in the risk index. Step 5 Deï¿½ine trigger rules (used in analysis to identify candidate actions [projects] for inclusion in the work plan). Step 6 Incorporate the risk index into the pavement management beneï¿½it calculations. Step 7 Identify the scenarios to be analyzed (such as funding constraints), as well as the speciï¿½ic set of assets (i.e., the scope) of the analysis. Step 8 Run analyses for the scenarios deï¿½ined in the PMS, compile results, and report projections. Table 10-1. Methodology for conducting asset-level risk index studies.
Protocols Developed for Studies That Were Not Piloted 109Â Â In this case, a large-scale vulnerability assessment that included culvert locations had been conducted and the data were provided for this study. A small subset of culvert locations was used to explain the proposed technique to develop an asset-level risk index. Additionally, NCDOT representatives provided rough cost estimates for culvert replacement and repairs. They also supplied estimates for how long a road would be closed for repair or replacement after a washout. Action Taken in the PMS Similar to the piloted Study 4, if this study were implemented in a PMS, the system would need to be configured to calculate the asset-level risk index for each pavement section. Decision trees and treatments would need to be updated with risk mitigation actions and the resulting reduction of risk from those actions. Finally, the optimization analysis would need to be configured to include risk in the objective function. For more detail, refer to Appendices G (Study 4) and I (Study 6). 10.1.4 Outputs from the Study As with the other asset-level risk index studies, the initial output from this study was a risk index indicating risk caused by non-NBIS culvert failure during storm events, at the asset level (road section) and summarized at the network level. The ideal outcome of implementation would be a set of optimized strategies for the culvert locations (or road sections) that shows if and when mitigation actions are to be taken to minimize the risk and/or consequences of washouts or flooding. These output strategies would be accompanied by long-term forecasts of risk level over time. FigureÂ 10-1 shows the output from the proof-of-concept study completed using the example NCDOT data. This figure exhibits two scenarios. The first scenario (solid line) shows the risk index forecasted for a scenario in which no funding was dedicated to mitigating risk. The sec- ond scenario (dotted line) shows an incremental decline in risk as risk mitigation projects were funded and selected by the simulated optimization. The forecasted change in the risk index depicted in FigureÂ 10-1 could be beneficial to an agency in building incentive to undertake projects that mitigate risks. In a management system, an associated project work plan would be produced for each scenario. The associated project $- $10,000 $20,000 $30,000 $40,000 $50,000 $60,000 $70,000 $80,000 $90,000 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Ri sk In de x Year Total Risk - Do Nothing Total Risk - Mitigation Figure 10-1. Network-level culvert washout risk projectionâ comparison of do nothing and mitigation scenarios.
110 Risk Assessment Techniques for Transportation Asset Management: Conduct of Research work plan for the risk mitigation scenario could be used to select and plan risk-mitigating projects. The detailed process for producing FigureÂ 10-1 is outlined in Appendix H. Overall, the proof-of-concept study shows the potential for existing pavement or general asset management systems to analyze and provide information that can be used to manage risk. This is accomplished by identifying optimum strategies for minimizing over an analysis period the risk resulting from threats and vulnerabilities identified in this study. 10.1.5 Participating DOT Organizational Unit(s) If completed in an existing PMS, the implementation of this method would be conducted by the pavement management group. There would need to be coordination with the agency group or groups responsible for programming and project selection decisions for maintenance, rehabilita- tion, and replacement of small (non-NBIS) culverts. This would typically be the central and field staff in the maintenance or bridge area in the agency. The study would also involve active participa- tion from additional groups, such as a hydraulics unit within the agency that might have conducted any applicable vulnerability studies. 10.1.6 Who in a DOT Could Use the Results and How The results of this study could be used by the group responsible for small non-NBIS cul- verts within the agency (likely the maintenance or bridge areas). This group could use results to develop recommendations for the management of risks to non-NBIS culverts. The study results could also be useful to planning departments and anyone conducting future vulnerability assessments. The study could provide supporting evidence for identifying the spe- cific assessment data that need to be captured as part of future vulnerability assessments. For example, the vulnerability of a culvert to washout may be known for a 10-year storm but not for a 100-year storm. As storm intensity and frequency increase, an agency may decide that vulner- ability studies are needed for 100-year storms as well. Agencies could use this methodology to incorporate culvert washout and flooding risk man- agement into project and funding decisions. Additionally, the study could be useful to those involved in developing and implementing the TAMP and in integrating risk into asset man- agement decisions. With extreme weather becoming a national challenge, this methodology could be helpful to agencies in conducting, planning for, and budgeting for projects to improve network resiliency. 10.1.7 Challenges to Expect and Suggestions on How to Plan for Them The difficulty in setting up and using this study would depend on the available data and the man- agement system used. For agencies that have a PMS in place, the additional setup and configuration to include small culvert risk would be relatively easy. If the agency has a general asset management system (where individual asset deterioration for the culverts could be modeled), the setup for this study would depend more on the effort spent to define the input data and the risk index. In addition to the configuration and depending on available data, identifying and defining initial threats, threat probabilities, vulnerabilities, and consequences may involve moderate effort. 10.1.8 Resources Needed by a DOT to Implement the Study Any development of a risk index for non-NBIS culverts pre-supposes that the agency has or will have available data regarding locations and other attributes of these assets. As with other
Protocols Developed for Studies That Were Not Piloted 111Â Â studies, the agency would have to commit some resource hours from the personnel who run the asset management system. Those personnel would need to have access to their management system as well as be able to perform the necessary additional configuration, run analyses, and generate reports of the results. In addition, some work could be required on the database to set up a specific test environment in which to conduct the study. Agencies will have different data available for use in this study and may replace assumptions with real data where available. The four components needed at minimum to complete this study are 1. GIS pavement section layer 2. Culvert location data, including culvert attributes to enable some form of vulnerability and consequence estimation (typically in the form of an adequacy or vulnerability type of data assessment) 3. Estimated mitigation costs 4. Configurable PMS in place In addition to the technical resources, the agency needs to assign an implementation cham- pion to ensure that the implementation provides the greatest benefit to the agency. The champion could be an agency SME who can guide the configuration, including the identification and quan- tification of initial threats, threat probabilities, vulnerabilities, consequences, mitigation actions, costs, decision rules, and analysis parameters. 10.2 Study 6: Asset-Level Risk Indexâ Landslide Hazard Management This study, though not piloted, illustrates how a risk index can be developed for geotechnical hazards along a roadway. These hazards could include rockfalls, rockslides, and debris flows, which pose a major threat to highway safety and mobility. Because these are often large-scale events, they are a threat to life and property beyond the roadway as well. The study was devel- oped to show how a landslide hazard risk index can be developed for use in an existing PMS. The general methodology presented here follows the asset-level risk approach also used in Studies 4 and 5. The scenario and guidance developed for testing include the following: â¢ Threats: Rockfalls, rockslides, and debris flows. â¢ Consequences: Roadway closed, serious injury, or fatal crashes. â¢ Implementation Effort: Define risk index input values for a subset of locations. Define how risk index fits into objective function. Run an analysis to optimize using a subset of locations, and analyze and summarize the resulting metric projections. 10.2.1 Study Objective and How the Results Help Inform Asset Risk Decisions The objective of this proof-of-concept study was to demonstrate a methodology for creating a risk-based index for managing the vulnerability of roadway slopes to slides of various types. Sim- ilar to the pilot study for pavement section flooding, this study was conducted at the asset level and illustrates the methodology for defining an asset-level risk index for use in a DOTâs PMS. A recent study for the West Virginia DOT Division of Highways (WVDOH) by Dr. Wael Zatar of Marshall University44 noted that in the United States, landslides occur in all 50 states, costing approximately $3.8 billion in damage and causing 25 to 50 fatalities annually. Dr. Zatar explains that slope saturation, the primary cause of landslides, can be caused by intense rainfall, snowmelt, changes in groundwater levels, and water-level changes along coastlines, earth dams, and the banks of lakes, reservoirs, canals, and rivers. Because landslides can cause severe traffic
112 Risk Assessment Techniques for Transportation Asset Management: Conduct of Research interruption to highway and road systems, it is beneficial to quantify that risk and determine where mitigation is necessary. 10.2.2 Description of the Technique or Tool This study developed a risk index based on a vulnerability study project performed for WVDOH on landslide hazard management. In this example approach, a risk index was cal- culated using a set of survey data provided by WVDOH pertaining to both the likelihood and consequence of a slope failure that may result in closure of the road while emergency repairs are carried out. This study provides interested agencies with information to implement similar studies to aug- ment decisions on rockslide or landslide risks. The formula used for calculating the risk index is provided in Appendix I. In addition, this example approach illustrates how this information could be integrated into a PMS and how a risk mitigation benefit could be modeled to occur. It is assumed that the landslide or rockslide mitigation would take place as part of various pavement projects. Like the other asset-level studies discussed in this report, the methodology in this study takes advantage of any existing vulnerability assessments on landslide hazards that the agency has already conducted or is in the process of conducting. The landslide hazard index previously devel- oped by Marshall University was used in this proof of concept. This study exhibits how the existing assessment could be incorporated into an existing PMS by aggregating the risk index values into a section landslide hazard risk index for all landslide hazard locations on a pavement section. Alternatively, the study exhibits how the risk index values could be integrated into a general asset management system in which each landslide hazard location would represent an individual asset. Once incorporated into the management system, the landslide risk index for each location could be used to select projects that address the mitigation based on benefitâcost analysis. 10.2.3 Methodology Used in Conducting the Study The methodology detailed in TableÂ 10-2 shows the same eight steps as Studies 4 and 5. This section provides a summary of the process while more details are outlined in Appendix I. The steps detailed in TableÂ 10-2 were followed with example data from WVDOH to simulate what an implementation of the methodology would entail. Input Data Used in the Study This study used information from an existing landslide hazard vulnerability study conducted for WVDOH and illustrated how the results could be integrated into the WVDOH PMS. The example data used the landslide hazard scoring formula developed as part of the vulnerability study. For this study example, the landslide hazard locations listed in TableÂ 10-3 were selected. Action Taken in the PMS Similar to the piloted Study 4, if this study were implemented in a PMS, the system would need to be configured to calculate the asset-level risk index for each pavement section. In this case, the risk index would be calculated using landslide hazards as threats. Decision trees and treatments would need to be updated with risk mitigation actions and the resulting reduction of risk from those actions. Finally, the optimization analysis would need to be configured to include risk in the objective function. For more detail, refer to Appendix G for this study and Appendix I where action was completed for the similar Study 4 in an actual PMS.
Protocols Developed for Studies That Were Not Piloted 113Â Â 10.2.4 Outputs from the Study The output from this proof of concept is an example landslide hazard risk index that would be stored for each pavement management section. This risk index could be forecasted along with the other pavement indices. The output also illustrates calculated benefits and costs of mitigation actions. Example forecasted optimization results were produced for the proof of concept. After analysis, the output from a PMS is expected to be a set of optimized project work plan strategies for the landslide hazard locations (or pavement sections). These strategies would show if and when mitigation actions need to be taken to minimize the risk of landslide hazards and associ- ated road closure delays, given a certain level of funding. Output strategies are accompanied by long-term predictions of future risk levels. FigureÂ 10-2 shows the output from the proof of concept completed using the example WVDOH data. This figure exhibits two scenarios. The first scenario (solid line) shows the risk index forecast for a Location ID Route County Milepost WV-42â001 WV-42 Tucker 1.1 WV-93â015 WV-93 Tucker 3.4 WV-28â003 WV-28 Tucker 2.5 WV-90â001 WV-90 Tucker 11.6 WV-560 003 WV-560 Tucker 7.2 WV-38â â 002 WV-38 Tucker 1.9 WV-72â001 WV-72 Tucker 0.3 WV-46â004 WV-46 Tucker 4.1 Sï¿½ï¿½ï¿½ï¿½ï¿½: WVDOH. Table 10-3. Example landslide locations used to illustrate the method for conducting asset-level risk index studies. Methodology Steps Step 1 Identify available sources within the organization of vulnerability studies that have been conducted for the identiï¿½ied threat or hazard type. Identify whether quantiï¿½ied threat probabilities, vulnerabilities, and consequences are available from an existing assessment. Deï¿½ine values for these for each vulnerable location within the study scope. Step 2 Determine and conï¿½igure in the management system the calculation of the quantiï¿½ied risk index component as a function of the threats, vulnerabilities, and consequences. Step 3 Decide on any deterioration models for changes in the threat probability, vulnerability, and consequences over time, possibly as the result of climate change. (Note that as- sumed deterioration rates can be constant.) Step 4 Identify risk mitigation actions and the associated reduction in the risk index. Step 5 Deï¿½ine trigger rules (used in analysis to identify candidate projects for inclusion in the work plan). Step 6 Incorporate the risk index into the pavement management beneï¿½it calculations. Step 7 Identify scenarios to be analyzed (such as funding constraints), as well as the speciï¿½ic set of assets (i.e., the scope) of the analysis. Step 8 Run analyses based on the two scenarios in the PMS, compile results, and report fore- casts. Table 10-2. Methodology for conducting asset-level risk index studies.
114 Risk Assessment Techniques for Transportation Asset Management: Conduct of Research scenario for which no funding was dedicated to mitigate risks. The second scenario (dotted line) shows an incremental decline in risk as mitigation projects are funded and selected by the simulated optimization. The forecasted change in the risk index depicted in FigureÂ 10-2 could be beneficial to an agency building incentive to undertake projects that mitigate risks. In a management system, an associated project work plan would be produced for each scenario. The associated project work plan for the risk mitigation scenario could be used to select and plan risk-mitigating projects. The detailed process for producing FigureÂ 10-2 is outlined in Appendix I. 10.2.5 Participating DOT Organizational Unit(s) This strategy would be implemented by the pavement management group in coordination with the agency group responsible for landslide hazards. The additional units involved may vary across agencies, though in many agencies it is likely to be the field maintenance offices. The implementation of this strategy could involve additional groups within the agency, such as the geotechnical engineering unit that might have conducted any applicable vulnerability studies. 10.2.6 Who in a DOT Could Use the Results and How The results of this analysis could be used by the group and decision makers responsible for landslide hazards management within the agency. With the reduction of highway fatalities being a high priority in every transportation agency, the areas in the agency involved in program and project prioritization could use the analysis results to augment decisions on safety projects. For states where such hazards are a high priority (for example, Vermont) the results could help an agency plan the short- and long-term management of rockfall and landslide hazards. Funding to mitigate such hazards requires planning. The results of such a study could be useful to anyone forecasting or requesting budgets to address forecasted risk levels. The information could also serve as input to the risk sections of the TAMP. The study results could also be useful to planning departments and anyone conducting future vulnerability assessments. The results can be used to provide guidance on the specific assessment data that need to be captured as part of future vulnerability assessments. 1,600 1,400 1,200 1,000 800 200 0 400 600 La nd sli de H az ar d Ri sk In de x Figure 10-2. Network-level landslide hazard risk index forecast under two scenarios.
Protocols Developed for Studies That Were Not Piloted 115Â Â 10.2.7 Challenges to Expect and Suggestions on How to Plan for Them The difficulty in setting up and using this study methodology would depend somewhat on the available data and the management system used. For agencies that have a PMS in place, the additional setup and configuration to include landslide hazard risk should be relatively easy. Similarly, if the agency has a general asset management system (where individual asset deteriora- tion is modeled), the setup for such a study would depend more on the effort put into defining the input data and the definition of the risk index. Depending on available data, the project scope, or both, identifying and defining initial threats, threat probabilities, vulnerabilities, and consequences may involve moderate effort. 10.2.8 Resources Needed by a DOT to Implement the Study Any development of a risk index for landslide hazards would require that the agency has or will have available data regarding locations and other attributes of these locations. If not, the agency will have to invest resources to collect needed data. As with other studies, the agency would typically need to commit some resource hours from the personnel who run the asset management system. The personnel would need to have access to their management system and be able to perform the necessary additional configuration, as well as be able to run analyses and report results. The implementation could also include setting up a database test environment in which the study can be conducted. Agencies will have different data available for use in this study and may replace assumptions with real data where available. The five components needed at minimum to complete this study are 1. GIS pavement section layer 2. Landslide location data 3. Landslide hazard locations and attributes to enable some form of vulnerability and con- sequence estimation (typically in the form of an adequacy or vulnerability type of data assessment) 4. Estimated mitigation costs 5. Configurable PMS in place In addition to the technical resources, the agency needs to assign an implementation cham- pion to engage the appropriate stakeholders. It would be ideal if the champion could also act as an SME to review the results for the adoption of the outputs and recommendations. Champions and SMEs need to be able to guide the configuration, including the identification and quanti- fication of initial threats, threat probabilities, vulnerabilities, consequences, mitigation actions, costs, decision rules, and analysis parameters. 10.3 Study 8: Program-Level Riskâ Bridge Network Analysis This study, though not piloted in an agency, illustrates a framework for evaluating funding risk at the bridge network or program level. In this proof-of-concept study, data from the City of Durham, North Carolina, were used to illustrate the technique. Detailed information in this section and AppendixÂ K is provided for use by agencies interested in implementing this study. This study can be implemented using any configurable BMS. The following are addressed in the study scenario: â¢ Threats: Unplanned changes to available funding levels, inflation rates, or deterioration rates â¢ Consequences: Impacts to forecasted condition metrics
116 Risk Assessment Techniques for Transportation Asset Management: Conduct of Research â¢ Implementation Effort: Define normal asset management system modeling inputs, including asset inventory, starting condition indexes, deterioration models, and action improvement models. Then analyze and summarize resulting metric projections. 10.3.1 Study Objective and How the Results Help Inform Asset Risk Decisions The objective of this study is to demonstrate an example analysis of program-level funding risk by forecasting bridge condition over time based on different funding levels. The results of this study would provide high-level information on bridge resources needed for decision making within an agency. In conjunction with other program-level data for other asset groups, these results could inform cross-asset trade-off decisions. 10.3.2 Description of the Technique or Tool This study could be implemented in an existing management system. The approach explored forecasting network-level bridge conditions over time for several scenarios involving the threat of variability in program-level input factors. These threats are the input variables selected for the study. This study proposed the use of existing BMSs or any asset management systems as a tool for assessing these threats. Examples of variables that can be used as input to the study include inflation rate, available funding levels, and adjusted deterioration models. 10.3.3 Methodology Used in Conducting the Study The methodology outlined in this study includes the two steps outlined in TableÂ 10-4. This section provides an overview of the process, input data, and expected output data covered in the study. The detailed steps and data involved in completing the proof of concept are available in AppendixÂ K. Input Data Used in the Study This study is based on data from a smaller agency, the City of Durham, North Carolina. The City of Durham does not have a fixed annual structures budget and needs to run analyses that look at how best to spend the overall available funds across multiple years. The base data available consisted of a population of 40 NBI bridges and culverts that are inspected regularly. Thus, inven- tory and condition data were available for the 40 bridges and used in this study. The scenario- level input data to the analysis consists of identifying the best-case, expected, and worst-case average funding levels per year. The funding levels selected for this study were â¢ Best-case funding level: $360,000 per year average â¢ Expected funding level: $317,000 per year average â¢ Worst-case funding level: $265,000 per year average Methodology Steps Step 1 Deï¿½ine which scenario input variables to consider as risk factors in the scenario. Determine levels of variation (best, worst, and expected cases) for input variables to use in the analysis. Step 2 Run analyses in the asset management system, analyze results, and re- port. Table 10-4. Methodology for analyzing bridge program risks.
Protocols Developed for Studies That Were Not Piloted 117Â Â Actions Taken in the BMS For the proof-of-concept study, the existing BMS was already configured to produce a Pareto- efficient frontier (explained in the following subsection). Therefore, the actions taken in the BMS would be minimal if inventory and condition data are up-to-date. For the study, the results of an analysis from the City of Durham were provided. If this study were implemented in an agency with a BMS not capable of producing an efficient frontier, multiple scenarios with different values used as input variables could be completed to produce a similar envelope of forecasted condition. Analysis of the impacts of changing vari- ables, including available funding levels, should be possible in most existing BMSs. 10.3.4 Outputs from the Study The general output from this study shows the average condition index for the three funding levels in a Pareto-efficient frontier format.45 This allows the agency to quickly identify the effects of potential drops in average funding as well as the effect of possible increased funding. A Pareto- efficient frontier is not a typical output from all BMSs. However, with some adaptation of the methodology, similar analyses of program-level variables can be performed in different BMSs. A similar analysis of program-level variables is also presented in Study 7. Also, as part of the study, specific projections are shown of the deck, superstructure, substruc- ture, and culvert NBI ratings for three specific scenarios: expected, best-case, and worst-case average funding levels. FigureÂ 10-3 shows the example results for the average substructure condi- tion for three scenarios selected from a Pareto-efficient frontier. Additional results in a format similar to FigureÂ 10-3 were reported for the superstructures and deck. Refer to AppendixÂ K for the detailed results. From all the results, it was noted that the cul- verts, decks, and superstructures could be maintained at or above current condition levels for all funding. However, as shown in FigureÂ 10-3, the substructures had worsened in condition by the end of the 35-year analysis period in all three funding scenarios. This indicates that even under the optimistic funding level considered in this study, the substructure condition was trending 7.8 7.9 8 8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Year Average of Expected Case - Substructure Index Average of Best Case - Substructure Index Average of Worst Case - Substructure Index Av er ag e Co nd iti on (N BI In de x) Figure 10-3. Network-level NBI condition forecasted for the substructure under the expected, best-case, and worst-case funding scenarios.
118 Risk Assessment Techniques for Transportation Asset Management: Conduct of Research downward. If these were actual funding values used by the agency, they would show that the funding allocated may not be sufficient to maintain condition over time. 10.3.5 Participating DOT Organizational Unit(s) The implementation of this study would involve coordination between those within an agency responsible for maintenance, rehabilitation, and replacement of bridges and large NBIS culverts. This study could involve agency personnel and decision makers responsible for funding deci- sions for the structures program. Financial staff could provide estimates for variation in future funding and inflation rates. Bridge management staff would be best suited to provide estimated variation in deterioration rates. 10.3.6 Who in a DOT Could Use the Results and How The results would be useful to bridge managers to build a case for requesting additional fund- ing or maintaining the current funding level. Additionally, anyone using forecasted budgets and resulting bridge conditions, including those considering risk in the compilation of a TAMP, would find these results useful. Being able to show results in terms of the best case, expected case, and worst case of funding needed annually allows decision makers to see the âenvelope,â or range of possible outcomes, based on the uncertainty of variables. This is beneficial for making cross-asset trade-off deci- sions. While this study example looks at uncertainty in funding levels, it would also be possible to similarly analyze other inputs such as inflation rate or deterioration rate. Agencies could also define scenarios with a combination of variable inputs. 10.3.7 Challenges to Expect and Suggestions on How to Plan for Them Agencies with a functioning BMS will not find it difficult to identify different scenarios using predicted ranges of input values. Although, depending on how many scenarios are identified for analysis, the setup and analysis of these scenarios may require some effort. The effort involved in the download, analysis, and compilation of results would also depend on the number of sce- narios identified. 10.3.8 Resources Needed by a DOT to Implement the Study To evaluate full Pareto frontiers as shown in this proof-of-concept study, the agency would need a functioning BMS capable of performing Pareto frontier multi-objective optimization to already be set up and in place within the agency. However, a similar analysis can be completed in any configurable BMS or asset management system to forecast the impacts of changing budget levels on network average conditions. The data and models in the system used would need to be up-to-date for the study to be most effective. The agency would typically need to commit resource hours from the personnel who have access to its BMS and are able to perform the necessary set up of multiple scenarios. They should be able to run analyses of these scenarios in the BMS and report results. Additionally, the agency personnel or SMEs involved need to understand scenario development and guide the definition of the scenarios and associated levels of input parameters to determine the applicable best-case, expected, and worst-case scenarios over the modeling period.
Protocols Developed for Studies That Were Not Piloted 119Â Â 10.4 Study 10: Decision Tree for Selecting Climate Risk Management Strategies Study 10 provides a decision tree for pavements impacted by higher temperatures to provide guidance on how to ultimately move from risk assessment to risk management of such pave- ments. Users could also apply similar decision trees to other asset and hazard combinations. The scenario and guidance developed for this study include the following: â¢ Threats: Rising average and extreme temperatures â¢ Consequences: Pavement deterioration or damage and roadway service disruptions â¢ Implementation Effort: Use the example decision tree study to inform decision making 10.4.1 Study Objective and How the Results Help Inform Asset Risk Decisions Study 10 provides a simple decision tree to explain the process of selecting risk management strategies to manage example climate risks. DOTs can use the decision tree to understand key considerations in choosing among varied adaptation options. 10.4.2 Description of the Technique or Tool Study 10 developed a decision tree for pavements subjected to higher temperatures. DOTs may use the decision tree to decide on the best approach to managing climate risks in upcoming projects as they enter the planning and design pipeline. For example, if the structure has a long remaining service life but is highly critical and already vulnerable, a replacement may be logical, whereas a structure with a short remaining service life at high risk from sea level rise in the com- ing decades could be a candidate for redesign upon natural replacement. The guidance walks DOTs through considerations, including asset lifetime, expected timing and severity of impacts, planning horizons, and available staff and funding resources, to choose from available risk management options. The types of risk management options include develop- ing and applying adaptive management techniques, accepting the risk, insuring against the risk, selecting operational tactics to manage disruptions, or engineering solutions to protect against the threat. FigureÂ 10-4 provides a decision tree for selecting climate risk management strategies for pave- ments expected to experience higher temperatures. Users could apply similar decision trees to other asset and hazard combinations. 10.4.3 Methodology Used in Conducting the Study TableÂ 10-5 details each step of the decision tree shown in FigureÂ 10-4 to help agencies imple- ment this guidance. Step 1: Consider Expected Asset Lifetime (Is the asset expected toÂ perform after mid-century?) In most areas of the country, after mid-century (around 2050) is when temperature condi- tions may begin to diverge significantly enough from historical conditions to require changes in pavement design and maintenance. (If temperatures in an agencyâs area are warming faster than average, the agency could consider an earlier time period.)
Is the asset expected to perform after mid-century? No action needed How critical is the road segment? Is current remaining useful life of the asset greater than 10 years? Critical or very critical Not so critical Is current remaining useful life of the asset greater than 10 years? Is this a high- volume segment? Given potential consequences and timing, are you willing to retrofit the project? Given the potential consequences, are you willing to accept the risk of inaction or insure against the risk? Accept or insure. See option (1) below. Use operational tactics. See option (2) below. ID tipping point. See option (5) below. Redesign. See option (4) below Retrofit. See option (3) below Given potential consequences and timing, are you willing to redesign? Given potential consequences and timing, are you willing to use operational tactics to manage the risk? Given potential consequences and timing, could you identify a tipping point to address at a later time? Yes No No No No No No No No No Yes Yes Yes Yes Yes Yes Yes Yes Figure 10-4. Decision tree for climate risk management strategies for pavements expected to experience higher temperatures. Methodology Steps Step 1 Consider expected asset lifetime. Step 2 Consider asset criticality. Step 3 Consider potential consequences versus risk tolerance. Step 4 Prioritize adaptation strategies according to risk tolerance. Table 10-5. Methodology for working through a decision tree.
Protocols Developed for Studies That Were Not Piloted 121Â Â Step 2: Consider Asset Criticality (How critical is the road segment?) All agencies define criticality differently but typically consider factors such as economic and social importance (e.g., passenger and freight traffic volumes, whether the asset provides access to essential destinations like economic centers, hospitals, or other emergency resources). See FHWA (2011) Assessing Criticality in Transportation Adaptation Planning for additional guidance.46 Step 3: Consider Potential Consequences Versus Risk Tolerance To determine potential consequences, agencies should consider projected temperature changes in their region and approximate the potential consequence of those changes in terms of poten- tial infrastructure damage costs and service disruptions. Known consequences of past events can be one input to this analysis. TableÂ 10-6 also summarizes some general impacts of changing temperatures on pavements. Step 4: Prioritize Adaptation Strategies According to Risk Tolerance The example options (1â5) incorporated into the decision tree are 1. Accept the risk and/or insure against the risk. â Description: Accepting the risk is similar to inaction, but it requires understanding the potential consequences of accepting the risk or taking action to assume the risk (e.g., through insuring against it). This adaptation strategy is most appropriate for low-volume road segments that are not deemed critical and have a remaining useful life of less than 10Â years. For pavements exposed to higher temperatures, this can mean accepting conse- quences such as the following: â¾ Physical deterioration: Premature deterioration could include conditions such as soft- ening of asphalt, embrittling, buckling, and rutting, migration of liquid asphalt, or subsidence. Thermal expansion can degrade concrete joints, steel, and asphalt. â¾ Consequences: These physical impacts can result in consequences such as overheated vehicles, tire degradation, and increased risk of tire blowout. Ultimately, degradation of pavements can result in possible short-term loss of public access or increased con- gestion of sections of road and highway during repair and replacement. These con- sequences can, in turn, affect traffic speed, travel time delay, crash risk, road closure, roadway capacity, and speed variation. â Limitations: This approach is not recommended for critical, high-volume, or long-lived roadway segments that may be retrofit, redesigned, or maintained or operated differently to reduce damage. 2. Use operational and maintenance tactics to manage risk. â Description: Operational tactics are appropriate for adaptation response when either (1)Â the asset has substantial remaining useful life (i.e., greater than 10Â years), but retrofits and/or redesign are not feasible, or (2) the asset has a shorter remaining useful life, but accepting the risk presents too great of a risk. Operational tactics include traffic manage- ment, monitoring of asset condition, changes to the timing of maintenance, and other management procedures to limit the impact. This approach does not impose additional capital or engineering design costs. â¾ Example: Many rutting problems are the result of pavement failure from the volume of transportation of heavy goods. If possible, consider rerouting this traffic. â¾ Example: Road deterioration will result in a greater need for maintenance of roads and pavement. Maintenance and construction costs for roads and bridges are likely to increase as temperatures increase. During repair and replacement, disruption can be managed via detours. â Limitations: This approach is not recommended for critical, high-volume, or long-lived roadway segments that may be retrofit, redesigned, or maintained or operated differently to reduce damage.
122 Risk Assessment Techniques for Transportation Asset Management: Conduct of Research 3. Retrofit â Description: Retrofitting involves making a physical adjustment to an existing design to accommodate projected changes in the design parameters. This strategy is an option in many cases for a threatened existing asset with multiple years of remaining useful life. â¾ Example: Sealants can be applied to asphalt surfaces to slow the oil oxidation process and extend the useful life of asphalt. Some sealants can also reduce surface temperature. â¾ Example: Concrete is most sensitive to high temperatures around joints, where con- crete can heave. Consider reinforcing joints. â Limitations: While retrofits such as protective cladding, coats, and sealants can help to extend the life of pavements, they are not able to reverse damage. Therefore, rather than retrofits, state DOTs can consider redesigning critical, high-volume, or longer-lived assets. Climate Change Impact Affected Components and Strategies Higher Average Temperatures Flexible Pavement â¢ Increased maximum pavement temperature increases the potential for rutting and shoving, requiring more rut-resistant asphalt mix- tures â May require raising high-temperature asphalt binder grade, in- creasing the use of binder polymerization, and/or improving ag- gregate structure in asphalt mixes â Increased use of rut-resistant designs including thin rut-re- sistant surfaces â¢ Increased age hardening of asphalt binder â Use of binders that age more slowly â Expanded use of asphalt pavement preservation techniques to address binder aging Rigid Pavement â¢ Increased potential for concrete temperatureârelated curling (and associated stresses) and moisture warping â Greater consideration of concrete coefï¿½icient of thermal expan- sion and drying shrinkage â Incorporation of design elements to reduce damage from ther- mal effects, including shorter joint spacing, thicker slabs, less- rigid support, and enhanced load transfer Higher Extreme Maximum Temperature In addition to strategies listed above, â¢ Higher extreme temperature may impact construction scheduling, requiring work to more often be conducted at night â¢ If accompanied by drought, increased potential for subgrade shrink- age Flexible Pavement â¢ Increased potential for asphalt rutting and shoving during extreme heat waves â See strategies for higher average temperatures, but recognize that the historical basis for selecting binder grades may no longer be valid Rigid Pavement â¢ Increased risk of concrete pavement âblow upsâ from excessive slab expansion â Use shorter joint spacing in new design â Keep joints clean and, in extreme cases, install expansion joints in existing pavements Warmer Extreme Minimum Temperature â¢ For all pavements, expect the depth of frost to decrease, thus reduc- ing the risk of frost heave â¢ May be able to require less depth of frost protection â¢ In areas with permafrost (e.g., Alaska) signiï¿½icant melting is antici- pated, resulting in serious impacts on ride quality Flexible Pavement â¢ Warmer minimum pavement temperature may allow for raising the low-temperature asphalt binder grade requirement Table 10-6. Climate change adaptation and pavement design- temperature items.47
Protocols Developed for Studies That Were Not Piloted 123Â Â 4. Redesign. â Description: Redesign involves reconsideration of the negative impact of temperature on pavements. Though pavements are designed to withstand high temperatures, they may not be designed to withstand the conditions of the future. Redesign is most appropriate for road segments that are high volume, critical, and expected to have a remaining useful life of more than 10Â years. It would also be appropriate to update design standards for new projects. â¾ Example: Consider the need for protection from wider temperature ranges. What tem- peratures lead to enhanced pavement rutting and more rapid pavement replacement? How can the choices made during design consider these wider temperature ranges? For example, pavement materials, construction details, pavement foundations, and bearings and expansion joints may need to withstand higher temperature exceedance thresholds over the next decades. â¾ Example: Consider pavement binders designed to withstand specific temperature thresholds and other temperature-resistant materials. Factors such as stone volume, aggregate type, and sand type affect thermal expansion. â Limitations: Given the costs associated with redesign, this approach is not recommended for noncritical, low-volume, or short-lived assets, unless the risk assumed is thought to be intolerable. 5. Identify a âtipping pointâ at which a future action would be taken. â Description: A tipping point is a given point in the future at which an adaptation may become appropriate for use. The point can depend on the state of the asset, the availability of funds, or the availability of new information. This option could be an appropriate response when it is unclear whether a critical, high-volume road segment with a remaining useful life of greater than 10Â years will face deterioration before replacement. In that case, it would be useful to monitor the road segment until it reaches the determined tipping point (e.g., the first signs of deterioration), at which point another adaptation approach will be applied. â Limitations: This approach is recommended for assets with a long remaining useful life, given that it delays decision making. The adaptation option needs to be identified at the same time that the tipping point is defined. For example, at the first signs of deterioration, a protective sealant will be applied to the road segment. Input Data Used in the Study To complete the decision tree, an agency will gather information about a given project, such as the location, expected lifetime, expected capital and maintenance costs, average annual daily traffic (AADT), and whether the asset serves any critical destinations. The decision tree walks users through key considerations, such as â¢ Defining the expected lifetime of the project and expected climate change impacts over the projectâs lifetime (e.g., a highway project with a 30-year design life) â¢ Analyzing inundation mapping to show expected conditions over the design life â¢ Defining consequences of potential climate changes (e.g., people or trips disrupted, other consequences) â¢ Defining the overall risk of the potential changes (likelihood versus consequences) over time â¢ Defining the risk tolerance â¢ Comparing different timing of options TableÂ 10-6 may be a particularly useful reference when applying Step 3 (consider potential consequences versus risk tolerance). Other potentially useful resources for understanding consequences, risk tolerance, and adap- tation strategy options are â¢ NCHRP Report 750: Strategic Issues Facing Transportation, Volume 2: Climate Change, Extreme Weather Events, and the Highway System: Practitionerâs Guide and Research Report (2014)48
124 Risk Assessment Techniques for Transportation Asset Management: Conduct of Research â¢ FHWA Vulnerability Assessment Scoring Tool (VAST) (2015)49 â¢ FHWA Climate Change and Extreme Weather Vulnerability Assessment Framework (2012)50 â¢ FHWA Tech Brief on Coefficient of Thermal Expansion in Concrete (2011)51 â¢ World Bank Group. Integrating Climate Change into Road Asset Management (2017)52 10.4.4 Outputs from the Study Completing the decision tree will result in a recommended course of action for managing climate risk to a given project. The study provides tangible examples and considerations that an agency can use in making decisions (i.e., what to do and when). 10.4.5 Participating DOT Organizational Unit(s) DOTs need to engage staff focused on planning and design to work through the decision tree and determine a course of action for managing climate risk to a given project. 10.4.6 Who in a DOT Could Use the Study Results and How Asset managers and project planners can use the decision tree to identify options and under- stand how to manage risks. 10.4.7 Challenges to Expect and Suggestions on How to Plan for Them The decision tree included in this study is designed to be an easy-to-use tool for DOTs to think through asset criticality, consequences, and risk tolerance to determine an appropriate course of action for a given project. A DOT may have additional questions or considerations to assess. DOTs are encouraged to modify the provided decision tree as needed to meet their needs or the unique context of a given project. 10.4.8 Resources Needed by a DOT to Implement the Study DOTs will need to gather information about a given project to answer key questions in the decision tree, such as location, expected lifetime, expected capital and maintenance costs, AADT, and whether the asset serves any critical destinations. Some specific resources are provided under the section on input data. 10.5 Study 12: Probabilistic Decision Tree for Risk Assessment Several tried and tested off-the-shelf tools are available in the marketplace that DOTs can use to facilitate risk-based decision making. Each tool has associated pros and cons, and a DOT will need to carefully evaluate the pros and cons before selecting a tool for analysis. This study evaluates one example tool to demonstrate risk-based decision making for a pavement treatment option that could provide substantial savings to a state DOT. 10.5.1 Study Objective and How the Results Help Inform Asset Risk Decisions The objective of Study 12 was to use an off-the-shelf tool available for risk-based decision making and data available from DOTs. When faced with the prospect of deciding on a pavement
Protocols Developed for Studies That Were Not Piloted 125Â Â treatment option to be selected from multiple options, each with different costs and uncertainty in results, DOTs may find it helpful to have an easy-to-use analytical tool to evaluate the options, expected costs and benefits, and probabilities of success in order to arrive at the most cost- effective option. The tool used for this study was the decision tree. It was applied to the problem of deciding whether to use a chip seal treatment plan for a pavement during a 10-year TAMP period versus a more expensive, albeit a longer-life, treatment of resurfacing with thin asphalt overlays. Cost and pavement condition data from the Puerto Rico Highway and Transportation Authority (PRHTA) were used for the decision tree analysis, along with pertinent data from other state DOTs. The data from PRHTA were chosen because PRHTA currently does not perform chip seal treatment. An evaluation of available pavement performance data (summarized in TableÂ 10-7) showed that nearly 29Â percent of the 8,224 lane miles of non-NHS lane miles managed by PRHTA (approximately 2,379 lane miles) are eligible candidates for chip seal treatment. At an estimated savings of approximately $36,000 per lane mile over a 10-year period, performing chip seal treat- ment could translate to approximately $86 million in savings. This is a substantial amount for the cash-strapped agency, making the current analysis a relevant one for PRHTA. Because the analysis incorporates probabilities for success of uncertain chance events, the output represents a risk-based analysis. 10.5.2 Description of the Strategy or Tool This study tests and illustrates the use of an Excel-based tool to develop a decision tree to decide between the implementation of a chip seal treatment plan or a resurfacing treatment for pavements in which uncertainties related to chance outcomes are involved. These uncertainties include success or failure of chip seal training and success or failure of chip seal treatment. A decision tree is used as a visual and analytical decision support tool in which the expected values of competing alternatives are calculated. The decision tree uses a graphical model to dis- play decisions and their possible consequences, including chance event outcomes, costs, and benefits. It has the structure of a flowchart in which each internal node (âchance nodeâ) repre- sents a test on an attribute (e.g., whether that attribute succeeds or fails), each branch represents the outcome of the test based on an assigned probability, and each decision node represents a decision taken after all attributes are computed. International Roughness Index Value < PRHTA Chip Seal Candidates Criteria 235 Rutting < Cracking < Eligible Lane Miles (Meets Criteria Above) Total Lane Miles % Eligible Total non-NHS lane miles under PRHTA Potential eligibility in lane miles for chip seal treatment 2,379 8,224 28.9% 3,594 1,039 10 0.3 Table 10-7. Computation of eligible PRHTA chip seal candidates based on analysis of available pavement performance data.
126 Risk Assessment Techniques for Transportation Asset Management: Conduct of Research A decision tree consists of three types of nodes: â¢ Decision nodes, typically represented by squares â¢ Chance nodes, typically represented by circles â¢ Terminal or end nodes, typically represented by triangles In the example tested, treatment-related information (e.g., cost, life, prerequisites, possible outcomes of potential decisions, and related probabilities) was used as inputs. The key decision analyzed was whether to perform a chip seal treatment or a resurfacing treatment. The logic used in developing the decision tree is described in greater detail in the methodology section that follows. The decision tree developed identifies the most optimal branch in the deci- sion node. For each branch, the probability-adjusted cost (prorated based on the probabilities and corresponding costs assigned to each branch, which represents a chance outcome following a decision) is computed. If more than one branch follows from a node, then the probability- adjusted cost is computed for each branch following the decision node, the costs are compared, and the branch with the least cost is identified in the decision node. The probability-adjusted costs for all outcomes emanating from the decision and chance nodes are computed from the total cost of the path back to the first decision node, and the least costly path is identified as the most optimal. A decision tree can be developed in a spreadsheet such as Excel, and the level of complexity desired can be incorporated. However, several free (open source) and inexpensive, easy-to-use software options, including add-ons to Excel, are available in the marketplace. If an agency desires to use decision tree analyses, they may choose an off-the-shelf option that best serves their requirements. 10.5.3 Methodology Used in Conducting the Study The broad logic (Step 3 in TableÂ 10-8) for the decision tree analysis used in the pilot testing is as follows. Based on the detailed FHWA guidance53 on performing chip seal treatment, worker training is an important prerequisite for the success and effectiveness of any chip seal treatment effort. If a chip seal treatment option is selected, then the first step is providing training. Then, depending on the success or failure of the training (when chance outcomes with probabilities of success or failure are involved), the next decision would be to either proceed with chip seal application (if the training is successful) or, alternatively, attempt to retrain the workers. Because of the estimated cost-effectiveness of chip seal treatment and the relatively low expense of training, a retrain option was included in the test. A DOT may choose to skip this step and move to resurfacing in the event of an unsuccessful training. Because a typical chip seal treatment application has an expected life of 5Â years, a second treat- ment would be necessary halfway through the TAMP period of 10Â years if a chip seal option is chosen. If the first chip seal treatment is successful, then a second treatment can be applied. On the other hand, if the first chip seal treatment fails, then it is assumed that a resurfacing treatment would be applied. As detailed in the 2019 PRHTA TAMP, pavement resurfacing at PRHTA has an expected service life of 14Â years,54 which extends beyond the 10-year TAMP period. Even though the cost incurred for resurfacing will be representative of its longer life as compared with the chip seal treatment, a prorated 10-year cost has been used for resurfacing in the analysis performed, to enable an apples-to-apples comparison. For the current testing, no future cost adjustment for inflation was incorporated into the analysis. The success or failure of the first chip seal treatment is also considered as a chance event with probabilities assigned to it.
Protocols Developed for Studies That Were Not Piloted 127Â Â If the retraining is successful, the logic used is retained for subsequent events. If the retraining fails, it is assumed that resurfacing is implemented. The success or failure of retraining is also considered a chance event. If the initial decision is to not provide worker training for chip seal treatment, then the assump- tion made is that chip seal will not be pursued as an option, and the decision is to implement resurfacing as the treatment. For each of the decisions and chance events, appropriate branches and corresponding probability-adjusted costs are computed. A step-by-step listing of the methodology followed is contained in TableÂ 10-8. Step 1 and Step 2: Obtain and Evaluate Data Input data for the study were collected and evaluated (Step 1 and Step 2). For the pilot testing, the input data included the cost of resurfacing using thin asphalt overlays in Puerto Rico, along with data relating to the expected life of a resurfacing treatment. The data relating to life expec- tancy of thin overlays for resurfacing were available from PRHTAâs 2019 TAMP. Because PRHTA has not implemented chip seal treatment, cost data from five state DOTs were collected for chip seal treatment, and treatment cost represented as a percentage of the overlay cost for those DOTs. The collected data were evaluated, and the average ratio of the chip seal cost to the resurfacing (ii) Success or failure of chip seal retraining [it was assumed that the probability of success for retraining would be higher than that for the initial training in Step 4(i)] (iii) Success or failure of chip seal treatment Step 5 Build a decision tree with the appropriate decision nodes and chance nodes using the logic established for the decision tree in Step 3. Step 6 Enter the costs and probabilities associated with each decision or chance option into the decision tree. Step 7 Methodology Steps Step 1 Obtain input data. Step 2 Evaluate the data to correlate the cost data from other states to determine the applicable chip seal costs in the studied location (Puerto Rico in the pilot study). For decision trees involving other parameters, complete any necessary data evaluation applicable to that parameter. Step 3 Finalize the overall logic for the decision tree. The logic used for the pilot testing is described in the introduction to this methodology section. Step 4 Based on available data and discussions with SMEs, estimate the proba- bility of success and failure of the chance events identiï¿½ied. For the pilot testing, the chance events identiï¿½ied were (i) Success or failure of chip seal training If an off-the-shelf tool is used for building the decision tree, the probability- adjusted costs and conclusions relating to optimal branches would be computed by the tool, indicating the lowest-cost option. However, if the decision tree is built from ï¿½irst principles, then the appropriate probability- adjusted costs for each branch need to be computed and the priority branches identiï¿½ied in the decision node. Table 10-8. Methodology for preparing decision trees.
128 Risk Assessment Techniques for Transportation Asset Management: Conduct of Research cost based on the data from five state DOTs was computed. It was assumed that this ratio would be reasonably representative for PRHTA. This ratio was therefore used to estimate the chip seal treatment cost in Puerto Rico based on the known PRHTA resurfacing costs. TableÂ 10-9 sum- marizes PRHTAâs cost of resurfacing, and the data from five DOTs used to estimate the ratio of chip seal cost to resurfacing cost are shown in TableÂ 10-10. As seen from TableÂ 10-10, the chip seal cost is on average approximately 22Â percent of the resurfacing cost. Applying this factor to the PRHTA resurfacing cost of $133,082 per lane mile (TableÂ 10-9), the estimated cost of one chip seal application is $29,476 per lane mile (approxi- mately 22.15Â percent of $133,082). The prorated 10-year cost of resurfacing works out to $95,059 ($133,082 Ã 10 Ã· 14). These two key cost figures were used in the decision tree analysis. It was also assumed that the cost of providing training is $1,000 per lane mile. Following the completion of Steps 1 and 2 above, Step 3 is initiated. The overall logic is repre- sented in FigureÂ 10-5. Ohio DOT Ohio DOT. Transportation Basics: Chip Seal55 âChip seals cost one-fourth to one-fifth the cost of asphalt resurfacing.â 20% âChip seals extend pavement life by five to seven years before a full resur- facing.â Washington State DOT Washington State DOT. Preserving Our Roads and Bridges: Chip Seal56 âThe cost of chip seals is 15%â20% of the cost of pavement overlays.â 20% South Carolina DOT 2019 TAMP57 Adds 4â6 years of pavement life Chip seal cost per lane mile $29,705 Rehabilitation cost per lane mile $147,949 Chip seal cost as % of rehab cost 20% Kentucky Transportation Cabinet 2019 TAMP58 Chip seal cost per lane mile $26,000 Thin overlay cost per lane mile $75,000 Chip seal cost as % of thin overlay cost 35% Indiana DOT 2019 TAMP59 Chip seal cost per lane mile $12,000 Hot mix asphalt thin overlay per lane mile $75,000 Chip seal cost as % of thin overlay cost 16% Average Chip seal cost as % of thin overlay cost 22% Table 10-10. Chip seal and resurfacing (thin overlay) costs from five state DOTs. Preservation 2-in Cold Milling & Overlay $133,082 Minor Rehabilitation 5-in Cold Milling & Overlay $312,161 Major Rehabilitation Full Depth (8-in) Cold Milling & Overlay $365,690 Reconstruction Full Depth (8-in) + 6-in Base Replacement $382,540 PRHTA Cost of Pavement Treatments Table 10-9. Cost of PRHTAâs pavement treatment options showing cost of thin overlays used to estimate resurfacing costs for decision tree analysis.
Figure 10-5. Overall logic of the decision tree.
130 Risk Assessment Techniques for Transportation Asset Management: Conduct of Research Step 3: Finalize Logic for the Decision Tree The overall logic summarized in the introductory paragraphs of the methodology is depicted in FigureÂ 10-5. Note that cost data for the different decisions and alternatives are not yet entered into the tree. Step 4: Estimate the Probability of Success and Failure of the Chance Events In Step 4, the probabilities of occurrence of the different events immediately following the chance nodes in FigureÂ 10-5 are established based on available data or estimated based on input from SMEs. The input of SMEs is used to arrive at these probabilities. For the study, the following probabilities were used: â¢ Success or failure of chip seal training â Probability of success of chip seal training = 0.75 â Probability of failure of chip seal training = 0.25 â¢ Success or failure of chip seal retraining â Probability of success of chip seal retraining = 0.85 â Probability of failure of chip seal retraining = 0.15 It was assumed that the probability of success for retraining would be approximately 10Â per- cent higher than that for the initial training. â¢ Success or failure of chip seal treatment â Probability of success of chip seal treatment = 0.8 â Probability of failure of chip seal treatment = 0.2 The probabilities established in Step 4 are also depicted in FigureÂ 10-5. Steps 5, 6, and 7: Build the Decision Tree In Step 5 and Step 6, the probabilities and the cost data are entered. In Step 7, the various probability-adjusted costs of each branch are computed to complete the decision tree. The com- pleted decision tree is shown in FigureÂ 10-6 and discussed in greater detail in the next section on outputs. 10.5.4 Output from the Study The completed decision tree using the steps described in the methodology section is shown in FigureÂ 10-6. The key nodes in FigureÂ 10-6 are identified as follows: â¢ Decision node: square identified by A, B1, B2, B3, C1, C2, and so on. â¢ Chance node: circular â¢ End or terminal node: triangular FigureÂ 10-6 shows the cost and probability data entered into the decision tree in color-coded boxes. The green boxes contain the costs of the different alternatives per lane mile. The blue boxes on the right of the figure show the total cost of each branch per lane mile. These are computed by following that branch from the rightmost to the leftmost decision node and summing the costs in the green boxes. For example, in FigureÂ 10-6, the total cost in the uppermost blue box is a summation of the costs in the green boxes along the branch formed by decision nodes C1-B1-A. Similarly, the cost in the fourth blue box from the top, $126,535, is a summation of the costs in the green boxes along the branch following decision nodes D2-C3-B2-A. The brown boxes indicate the probabilities associated with each chance event that follows the chance node.
Figure 10-6. Completed decision tree relating to the chip seal versus resurfacing decision making for PRHTA.
132 Risk Assessment Techniques for Transportation Asset Management: Conduct of Research The numbers within the yellow decision node boxes identify the branch with the lowest cost that follows the decision node until the next decision node. The number 1 within the decision node references the uppermost of the alternatives. For example, if two branches are emanating from a decision node, the uppermost would be identified as 1 and the lower as 2. The gray boxes show the probability-adjusted costs as computed for each branch following a node. The costs are computed from the total cost computed (displayed in the blue boxes at the right of the figure) and work leftward following each branch to the primary decision node (NodeÂ A). A review of FigureÂ 10-6 shows that the first decision node on the left (Node A) shows the optimal branch as Branch 1 and shows the cost as $74,181, which is the lesser of the two options ($74,181 versus $95,059) and is derived as follows: $74,181 = $73,069 Ã 0.75 + $77,517 Ã 0.25 In other words, Branch 1 multiplies the probabilities associated with each following chance event and the corresponding cost computed for that event. Similarly, the âtraining successfulâ branch points to Branch 1 to its right (i.e., âapply chip seal 1â and âchip seal successfulâ) as the lowest-cost alternative of the two alternatives. Because the decision node to the right of the chance node associated with âtraining successfulâ has only one branch, the probability-adjusted cost of that branch ($73,069) is depicted for that decision node. The probability-adjusted cost of $73,069 is computed as follows: $73,069 = $59,952 Ã 0.8 + $125,535 Ã 0.2 Each of the two costs is derived from the costs of the branch to its right. This logic is used throughout the decision tree to compute the probability-adjusted costs for each branch. The least costly ($59,952) and therefore the most optimal branch is the one where chip seal training is successful on the first try followed by a successful application of the chip seal treat- ment. The next optimal branch is when the team is retrained and then the chip seal training and treatment application are successful ($60,952). The decision tree also shows that applying a second chip seal treatment or a resurfacing treatment after the failure of training followed by failure of the application of the first chip seal treatment is the costliest and least optimal of the alternatives ($125,535 and $126,535, respectively). 10.5.5 Participating DOT Organizational Unit(s) This analysis was coordinated with the PRHTA consultants contracted to manage the agencyâs pavement management processes. Their tasks include overseeing the collection of data, conduct- ing quality control of the data submitted to the HPMS, and developing a PMS. The results were also shared with the in-house PRHTA pavement management staff. Among the challenges faced in this analysis was that pavement condition data were not avail- able for the entire non-NHS system. This lack of data was overcome by using the available data and only estimating potential territory-wide benefits. Another challenge was a lack of data on chip seal costs and expected service life in Puerto Rico. This challenge was overcome by using nationwide data for variables such as the length of service life. One challenge that could not be overcome was the lack of ADT and AADT values. Conse- quently, those were excluded from the analysis.
Protocols Developed for Studies That Were Not Piloted 133Â Â 10.5.6 Who in a DOT Could Use the Study Results and How The study results will be considered by the PRHTA pavement staff and the pavement man- agement consultants as they consider options to present to the agencyâs executive leadership. At the time of this study, PRHTAâs finances were severely constrained, and PRHTA Interstate and non-Interstate NHS pavements were well below national averages. Saving money on the non- NHS pavements could free up funds to further improve the Interstates along with the non-NHS pavements. These analyses can be included among options presented to executive leadership for how to improve pavement conditions with the agencyâs limited resources. 10.5.7 Challenges in Pilot Study Setup Little difficulty was experienced in conducting this analysis, other than the need to rely on some national data in lieu of PRHTA cost data. Until PRHTA tries chip seals, the agency will lack localized cost and performance data. Initial analyses will need to rely on assumptions such as those used in this example. Each time PRHTA solicits a bid for a chip seal project, it could use the bid information to update the cost assumptions. For each year that a chip seal is in place, the agency can update its service-life assumptions. Analyses of when and where to apply chip seals would benefit from traffic-volume data. As PRHTA improves its pavement management processes, chip seal analysis could benefit from assigning ADTs and AADTs to each pavement section. Then, traffic-volume data could be added to the process of selecting candidate sections. Software to produce decision trees is available for as little as a few hundred dollars. The analyses were conducted using a simple add-in Excel tool. 10.5.8 Resources Needed by a DOT to Implement the Study Initially, the only resources needed would be an understanding of how to set up a decision tree and national data for costs and service lives. The Excel add-in was the only software needed. National cost and performance data were readily available from agenciesâ TAMPs or other sources cited in this study.