Application and Validation of Remaining Service Interval Framework for Pavements

CHAPTER 2. RSI ALGORITHM AND AGENCY SELECTION

INTRODUCTION

This chapter documents the algorithms and methodologies formulated to perform the RSI application and validation analysis at the project level, the network level for two selected State transportation departments, and the strategic level using the PHT analysis tool. This chapter also documents the collection and review of information pursued to select two State transportation departments with good PMS practices to provide data in support of the validation analysis and actively participate in the project.

The development of the initial detailed analysis methodologies to implement the RSI concept at the project, network, and strategic levels are also detailed in this chapter. In development of these methodologies, the project team expected to use existing agency construction triggers, threshold limits, and performance prediction models as the basis for the computation of the RSI numerics. The methodologies presented in this chapter are general in nature so that they are applicable to both agencies and the PHT analysis tool.

At the onset of the project, the RSI concept was to take into account the lifecycle cost (LCC) of the pavement system based on determining the RSI numerics of preservation, rehabilitation, and reconstruction as a function of agency thresholds for providing an acceptable level of service (LOS) and construction triggers. More specifically, the RSI concept was to be validated within the LCC framework to create a consistent construction event-based terminology and understanding (i.e., types of construction events and the timing of those events within the LCC concept, risk analyses, and other prioritization approaches based on streams of future construction events and benefits to facility users).

While refining the general computation algorithms presented in this chapter, the RSI validation approach evolved from simply demonstrating the RSI concept within existing PMSs to demonstrating the concept within an “ideal” PMS where decisionmaking considers the optimal treatment selection, not based on thresholds, but considering all possible treatments and treatment timings to select the optimal timing for treatment selection while maintaining an acceptable or above an acceptable LOS based on agency specifications.

As the project progressed, benefits to using the optimum timing when applying a treatment (i.e., preservation, rehabilitation, or reconstruction) to better represent the ideal PMS and better support performance management arose, and, as a result, optimum timing was used within the RSI validation methodology instead of thresholds. This evolved approach is demonstrated in chapters 4 and 5.

PROJECT-LEVEL ALGORITHMS USING LTPP DATA AND CalME

The project-level validation objective was to demonstrate how the inclusion of structural measurements in the selection of rehabilitation strategies is beneficial in selecting the optimal treatment sequence to yield the LLCC for a given pavement section. Project-level validation compared five different treatment scenarios based on CalME analysis, which determined the performance extension of the pavement for each treatment based on its structural and functional condition at the time of analysis. The equivalent uniform annual cost (EUAC) for each treatment was determined after taking into consideration what would be required to bring the pavement to a state of good repair in order to select the optimal treatment based on LCC considerations. The Secretary of Transportation defined a state of good repair as “a condition in which the existing physical assets, both individually and as a system (1) are functioning as designed within their useful service life, (2) are sustained through regular maintenance and replacement programs”⁽⁶⁾ (p. 2). The approach used in the project-level validation is detailed in chapter 4.

NETWORK-LEVEL ANALYSIS ALGORITHMS FOR STATE TRANSPORTATION DEPARTMENT PMSS

The initial approach to the network-level validation was to formulate a methodology for simulating the functionality of State transportation department PMSs over a 30- to 40-year analysis period based on the respective agency models. This goal would be accomplished by developing computer algorithms that mimicked pavement condition forecasting processes and the application of only LOS-based threshold limits using State transportation department procedures that produced LLCC. This process would theoretically generate a complete stream of future construction events for a network that the RSI numerics could be based on.

The 30- to 40-year analysis period was selected in order to show the long-range outcome of the network using combinations of preservation, rehabilitation, and reconstruction treatments in order to formulate the RSI numerics, which reflect the optimal treatments for LLCC. The extended analysis period was required to provide enough time for the optimal string of treatments to be selected in order to produce the LLCC.

Data and Pavement Performance Models

The network validation was to use State transportation department PMS information for the purpose of validating the RSI concept, including the following:

• Pavement condition and performance models (e.g., International Roughness Index (IRI), cracking, rutting, friction, and composite indices).
  
  
• Pavement condition threshold values used to trigger construction events, when identifying optimum time for the treatment is not possible due to lack of data or performance models.
  
  
• Construction cost information.
  
  
• Decision trees (e.g., what treatments are applicable based on condition or functional class).
  
  
• Rules, if any, applied within the State transportation department PMS (i.e., pavement sections with inadequate skid resistance are prioritized first prior to optimization due to safety concerns, etc.).
  
  
• Treatment benefits (e.g., if preservation treatment is placed when IRI = 120 inches/mi, what is IRI after treatment?).

No changes to the State transportation department PMS data or models were to be made in support of the RSI validation analysis effort.

Construction Events

While agencies may have multiple types of preservation or rehabilitation treatments, for planning purposes, they can be and generally are grouped under broader categories. For this effort, construction events were limited to the following four options:

• Do nothing.
• Pavement preservation.
• Rehabilitation.
• Reconstruction.

This simplification of the treatments used by agencies was recommended during the initial development stage of the RSI validation analysis process. The details of the actual treatments used within each category were to be implemented during the evaluation stage for each agency and are presented in chapter 5.

RSI Network-Level Validation Process

This section describes the process that was to be used in the network-level analysis based on the available data, pavement performance models, and construction events. The process described in this chapter shows the methodology that was to be applied to State transportation department data. The actual application of this methodology is presented in chapter 5.

The fundamental network-level approach was to develop alternative construction time histories allowing for various combinations of construction events. For each pavement section in the network, the different combination of construction time histories was determined based on the timing of each construction event, which could be threshold-driven or based on optimum timing and the improvement of the pavement condition as a function of the treatment timing. The pavement deterioration process then proceeded until the next construction event was triggered or applied. This process was continued until the pavement section under analysis reached the end of the analysis period.

A partial illustration of the hypothetical options that could result for a section in question is provided in table 1. A total of eight alternative construction time history options are shown. Each option shown in the top row is the combination of time history construction events shown in the subsequent column cells by year. Different road segments can have a different mixture of time history construction events to be evaluated. Once all the options were identified for a section, the optimal option was selected based on the LLCC.

This shows that the section can be treated at different times with various treatments. Preservation can be placed early, or reconstruction can be delayed to later years. It is important that the LLCC for the section is used to determine the optimal treatment when comparing these alternatives.

Analysis Outputs

Ultimately, the outputs for the network analyzed (based on optimal option for each segment) are shown in table 2 and table 3. Table 2 shows the RSI numerics for each segment. For example, for segment 1, the RSI numerics are preservation treatments in years 3 and 5, followed by a rehabilitation in year 18, and then preservation in year 22. A similar RSI was calculated for each segment. Table 3 illustrates the costs associated with the RSI numerics outputs from table 2 for all segments in the network. It should be noted that table 2 only shows a sample of the RSI output and not the entire network. For example, the total cost for year 1 were $1,240,000, $24,355,000, and $5,546,000 for preservation, rehabilitation, and reconstruction, respectively.

STATE TRANSPORTATION DEPARTMENT SELECTION FOR NETWORK-LEVEL VALIDATION

For network-level validation, two State transportation departments needed to actively participate in the project. A list of seven potential State transportation departments was developed along with the reason for these recommendations. The seven potential State transportation departments included the following:

• Colorado.
• Maryland.
• Michigan.
• Minnesota.
• North Carolina.
• Virginia.
• Washington.

The following criteria were used to evaluate the potential State transportation departments:

• Has an interest in participating in the study.
  
  
• Has developed construction triggers, threshold limits, performance curves, and strategy selection algorithms that can be used to develop RSI numerics.
  
  
• Would provide details on algorithms used to perform processes previously described in the RSI Network-Level Validation Process section.
  
  
• Would provide PMS data and PMS outputs.
  
  
• Would appoint a project liaison and will meet with the project team at least twice.

A summary of each potential State transportation department follows along with a summary of the selection of two agencies based on a comparison to the previously listed criteria and their suitability for participation in the project.

Colorado Department of Transportation (CDOT)

CDOT previously used the RSL concept as a measure of the State’s pavement quality. ⁽⁷⁾ However, in 2013, due to issues with the RSL-based system, CDOT implemented the drivability life (DL) analysis. The DL analysis aims to maximize acceptable driving conditions across the network and is a function of smoothness, pavement distress, and safety based on IRI, cracking, and rut depth data.

DL represents how long a highway will have acceptable driving conditions. The DL scale used by CDOT is as follows⁽⁷⁾:

• High DL: > 10 years.
• Moderate DL: 3–10 years.
• Low DL: ≤ 2 years.

CDOT uses the infrastructure asset management software to generate a list of resurfacing recommendations or strategies.⁽⁸⁾ This software determines the recommendations by optimizing the incremental benefit-cost ratio (BCR) to maximize the benefit for the network for the given budget.⁽⁸⁾ CDOT’s goal is to have 70 percent of the resurfacing projects match the recommendations from the PMS. They are currently at 76 percent.⁽⁹⁾

MDSHA

In 1997, MDSHA began developing and implementing a pavement management approach based on optimization. During an enhancement of the pavement management practices, MDSHA combined probabilistic and deterministic models for use in forecasting and planning analyses at the network and project levels, respectively.⁽¹⁰⁾ The probabilistic models were derived based on performance distributions for treatments, which were then used to develop the deterministic models in the form of performance curves.⁽¹⁰⁾

MDSHA also promotes a pavement preservation mentality. The pavement preservation program is based on a 6-year transportation program and relies on the PMS to develop performance-based pavement preservation plans.⁽¹¹⁾ MDSHA has developed a Pavement Preservation Guide that helps in the selection of the correct treatment through use of flowcharts, decision trees, and treatment tables.⁽¹²⁾ Through the use of this guide, a series of treatment options can be determined for further consideration using more specific project information.⁽¹²⁾

The core processes of the pavement management program are performance monitoring, model development, network optimization, project selection, funding approval, pavement design, and construction and maintenance.⁽¹¹⁾ MDSHA PMS services include data collection, processing, analysis, and improvements to the PMS. These services use Web-based tools for network-level optimization and project-level selection using BCR and RSL.

MDSHA uses a PMS software to optimize their budget planning and project selection. The software includes optimization tools (BCR, remaining life, etc.), asset management, geographic information systems, an image viewer, data warehousing, and data viewer tools.

Michigan Department of Transportation (MDOT)

MDOT has a mature PMS that was developed in the 1980s. MDOT also emphasizes a pavement selection strategy developed in the 1990s that consists of rehabilitation and reconstruction (R&R), capital preventative maintenance (CPM), and reactive maintenance. CPM is used to manage pavements with an RSL greater than 2 years, and R&R is used for pavements with an RSL of less than 2 years.⁽¹³⁾

MDOT evaluates the condition of the pavement systems using both a sufficiency rating and a PMS rating. The sufficiency rating consists of an annual subjective windshield survey, while the PMS rating consists of detailed pavement condition data collected biennially.⁽¹³⁾ Using the collected pavement condition data, such as cracking, raveling, flushing, spalling, faulting, roadway curvature, pavement grade, cross slopes, rutting, and ride quality, a Distress Index and a Ride Quality Index (RQI) are computed.⁽¹³⁾ MDOT considers a pavement with an RSL of zero and an index value of 50 or greater to be in need of reconstruction or major rehabilitation.⁽¹³⁾ MDOT has several performance measures for the trunkline (consisting of all State highways) network, including 90 percent of pavements in fair or good condition based on IRI according to FHWA standard and 90 percent of pavements having an RSL of 3 years or greater.^(14,13)

MDOT uses the Road Quality Forecasting System (RQFS) to evaluate pavement strategies for both short- and long-term condition levels by predicting future pavement condition and determining funding needs to meet desired conditions.⁽¹³⁾ The RQFS determines the percentage of the network that moves between RSL categories based on the suggested project selection and selects the strategy that is most effective and promotes a “preserve first” strategy.⁽¹⁵⁾

Minnesota Department of Transportation (MnDOT)

The MnDOT PMS contains an estimate of pavement RSL. As part of its annual report, the MnDOT Pavement Management Unit determines RSL for all highway segments. The RSL is determined as the number of years until an RQI of 2.5 is reached. The RQI is a smoothness index with a 0 to 5 scale; increasing values representing smoother roadways.

MnDOT implemented the Highway Pavement Management Application (HPMA) PMS in 1987.⁽¹⁶⁾ MnDOT uses HPMA in supporting the decisionmaking process through performance models, decision trees, and treatment selection; however, the MnDOT districts have significant influence on the project selection with input from the Pavement Management Unit, making it decentralized.⁽¹⁶⁾ The PMS considers preventative maintenance (crack seal/fill, rut fill, chip seal, thin non-structural overlay, concrete joint seal, and minor concrete repair), rehabilitation (medium overlay, thick overlay, medium mill and overlay, thick mill and overlay, and major concrete repair) or reconstruction (cold in-place recycling, rubblized portland cement concrete (PCC) and overlay, unbounded concrete overlay, and full-depth reclamation).⁽¹⁶⁾ MnDOT uses the performance curves in HPMA to predict RSL based on RQI. The RSL and future pavement condition are used to provide information regarding the impact of various funding scenarios.⁽¹⁷⁾

North Carolina Department of Transportation (NCDOT)

NCDOT uses a transparent, systematic, and data-driven process for prioritizing the major transportation components in the State and making investment decisions. This process, developed in collaboration with key partners, evaluates the benefits the project is expected to provide, the project’s multimodal characteristics, and how the project fits in with local priorities. NCDOT’s first Strategic Prioritization Process (known as Prioritization 1.0) was implemented in 2009 and was subsequently codified into law in 2012.⁽¹⁸⁾ NCDOT implemented the third generation of the Strategic Prioritization Process in 2014. NCDOT has also developed the Interstate Maintenance Preservation Program, a rating system for application of pavement preservation treatments that is unique and could be investigated with the RSI concept.

NCDOT’s PMS inputs the condition database, decisions trees, and performance models and then uses a multiobjective, multicriteria optimization analysis to output project-level lifecycle reports, network-level investments and funding strategies, forecasted conditions at the network and project levels, and comparative analyses of investment strategies.⁽¹⁹⁾

Virginia Department of Transportation (VDOT)

VDOT has a mature PMS that was developed in the 1980s. Currently, it uses the Critical Condition Index (CCI) to categorize pavement condition. VDOT aggregates pavement condition data collected by their vendors into load-related distress ratings (LDRs) and non-load-related distress ratings (NDRs). The lower of these two ratings is taken as the CCI. NDR considers transverse and longitudinal cracking, longitudinal joint separation, bleeding, etc., and LDR considers distresses caused by vehicle loads such as fatigue cracking, patching, and rutting.⁽²⁰⁾ Pavement condition is assigned according to the CCI value ranging from excellent to very poor, with a CCI rating of 60 or greater representing sufficient condition.⁽²⁰⁾ The statewide target is for 82 percent of interstates to be rated as sufficient. (As of 2012, 82.9 percent of interstates were rated as sufficient.⁽²⁰⁾⁾ VDOT also evaluates pavements based on IRI. Pavements are considered deficient in terms of ride quality if the IRI is 140 inches/mi or greater for interstates and primary roads or 220 inches/mi for secondary roads.⁽²⁰⁾ The statewide target is to have 85 percent of interstates with sufficient ride quality. (As of 2012, 93.3 percent of interstates have sufficient ride quality.⁽²⁰⁾⁾

VDOT implemented a new PMS in 2010 with features such as analysis of current pavement conditions, pavement performance modeling and forecasting, and calculation of performance-based needs expectations.⁽²¹⁾ The PMS also includes a multiyear optimization strategy selection tool, which considers the maintenance alternatives of do nothing, preventative maintenance, corrective maintenance, restorative maintenance, and major rehabilitation/reconstruction.⁽²¹⁾ The network-level analysis can base the optimization on maximizing the benefit, maximizing the condition indicator, or minimizing the total cost as a function of treatment costs or desired condition level.⁽²¹⁾

Washington State Department of Transportation (WSDOT)

WSDOT has a mature PMS, which was developed in the 1970s and fully implemented in 1982. It is considered a national leader in the field.⁽²²⁾ Based on the collected pavement distress data, WSDOT assigns three different pavement condition indices—pavement structural condition (PSC), pavement rutting condition (PRC), and pavement profile condition (PPC)—on a scale of 0 to 100, where 0 represents extensive distress and 100 represents a pavement with no distresses.⁽²²⁾ Using the PMS and performance curves, WSDOT projects when any one of the condition indices will reach 50 and determines the ideal time for rehabilitation using LLCC and other techniques to select optimum pavement construction strategies within a 6-year investment program.⁽²²⁾ Through this process, WSDOT promotes pavement preservation.

WSDOT uses WebWSPMS as the principle application for their pavement asset management. The Web-based WebWSPMS, which was developed in-house, provides an interface for accessing and viewing data from several sources including roadway configuration, location information, contract history, traffic information, capital projects, pavement activities completed by maintenance, construction contract costs and milestones, condition information, imagery, and data synthesis and analysis.

Agency Selection

Using the criteria set out at the beginning of the State Transportation Department Selection for Network-Level Validation section and the summary of the agencies provided, MDSHA and WSDOT were selected for participation in the study. Both agencies had developed construction triggers, threshold limits, performance curves, and strategy selection algorithms and used the RSL concept. In addition, both agencies were willing to provide details for the strategy selection algorithms and access to their PMS data and PMS outputs and to work with the project team to meet the objectives of the project.

STRATEGIC-LEVEL ANALYSIS ALGORITHMS FOR PHT ANALYSIS TOOL

In addition to project- and network-level validations, the proposed RSI concept was to be validated at the strategic level using FHWA’s PHT analysis tool. However, as detailed in chapter3, the models implemented into the current version of the PHT analysis tool did not support the validation of the RSI. The PHT analysis tool estimates pavement RSL for highway sections based on data items described in the 2010 revision of the HPMS database maintained by the FHWA.⁽²³⁾ The PHT analysis tool quantifies the RSL of the pavement for each highway section using the simplified American Associations of State and Highway Transportation Officials (AASHTO) Mechanistic-Empirical Pavement Design Guide (MEPDG) based performance prediction models.⁽²⁴⁾ The RSL is the number of years, or equivalent single-axle loads, remaining until pavement distress reaches a level where action is warranted.⁽⁴⁾

Proposed Algorithm for the Strategic-Level Validation

The validation of the RSI at the strategic level included adding an RSI module to enhance the PHT analysis tool. The RSI module was proposed to be developed as a plug-in to be integrated with the PHT analysis tool. The RSI module used the simplified pavement performance models (IRI, cracking, etc.) contained in the PHT analysis tool, which were based on the MEPDG.⁽²⁴⁾ No changes to the PHT analysis tool performance models were made.

The RSI module was designed so that for a given set of HPMS 2010+ data, analysis period, and minimum LOS to be maintained in addition to other constraints currently included in the PHT analysis tool, it provided a list of lifecycle cost analysis (LCCA) optimized series of treatments, timing, and costs for each pavement section over the analysis period. The remainder of this section will provide details of the algorithm used.

Setup and Rules

The proposed strategic level RSI algorithm was to consider the following four types of construction events:

• Do nothing.
• Pavement preservation.
• Rehabilitation.
• Reconstruction.

Each construction event had a threshold limit, which increased from preservation to reconstruction (i.e., the threshold limit of percent cracking increases from 15 to 25 percent from preservation to reconstruction). For example, table 4 from the Pavement Remaining Service Interval Implementation Guidelines can be used to define the threshold limits for reflection cracking or the IRI.⁽²⁵⁾

Each of these construction events also costs more to implement, with preservation being the least costly and reconstruction being the most costly. The algorithm was to abide by the following rules when considering the construction events:

• More than two successive preservation treatments were not allowed. Although this is an acceptable practice by many agencies and is often used, this rule was implemented to limit the choices for illustration purposes.
  
  
• Only one rehabilitation event within a 10-year period was allowed.
  
  
• Only one reconstruction event within a 20-year period was allowed.

Calculation of LLCC

In order to calculate the LLCC for a given highway section, the algorithm was to operate as follows:

1. For each pavement section, use PHT analysis tool prediction models to calculate the year when a pavement preservation, rehabilitation, or reconstruction treatment was needed based on the set thresholds. A threshold-driven approach was used because modification of the PHT analysis tool to fully incorporate an LCC-based approach was not feasible at the time of this project.
   
   
2. Create two construction scenarios (apply treatment from step 1 or do nothing) while considering the rules stated in the setup and rules.
   
   
3. For each construction scenario, determine associated construction costs and year when next construction event is triggered.
   
   
4. If the analysis period has not been reached, update model inputs and go back to step 1.
   
   
5. If the analysis period has been reached, consider all scenarios and find LLCC option for highway section.
   
   
6. Record LLCC option (year of treatments and costs) for the highway section and repeat steps 1 through 5 for other highway sections in network.

Figure 3 shows a flowchart highlighting the algorithm’s operation. The flowchart shows that for an available highway section, the prediction models are used to identify a year that triggers a construction event. If the analysis period is not yet met at that year, then two scenarios are considered: do nothing or apply treatment. This cycle is continued until the analysis period is reached for each highway section. The lowest LCC option from all possible scenarios is then selected and added to the summary table.

Figure 3. Flowchart. Proposed RSI algorithm.

Implementation Strategy

The PHT analysis tool has the capability to import HPMS 2010+ data. The PHT analysis tool pavement deterioration models are simplified MEPDG models similar to those developed for the Highway Economics Requirements System (HERS).⁽²⁶⁾ By using the PHT analysis tool, the proposed RSI algorithm’s implementation reused the input and prediction models steps of the basic RSL process in the PHT analysis tool. The proposed RSI implementation, therefore, focused on the strategy selections step of the RSI module while allowing PHT to take care of the input and the prediction model steps.

The following list includes the data requirements for the PHT analysis tool and the RSI algorithm from HPMS 2010+ data:

• AADT.^[1]
• AADT_Combination.
• AADT_Single_Unit.
• Base_Thickness.
• Base_Type.
• Begin_Point.
• Climate_Zone.
• County_Code.
• Cracking_Length.
• Cracking_Percent.
• End_Point.
• Expansion_Factor.
• Facility_Type.
• Faulting.
• F_System.^[2]
• Future_AADT.
• Future_AADT_Year.
• IRI.
• Lane_Width.
• Last_Overlay_Thickness.
• NHS.^[3]
• PSR.^[4]
• Route_ID.
• Rutting.
• Shoulder_Type.
• Soil_Type.
• Speed_Limit.
• State_Code.
• Structure_Type.
• Surface_Type.
• Thickness_Flexible.
• Thickness_Rigid.
• Through_Lanes.
• Urban_Code.
• Volume_Group.
• Year_Last_Construction.
• Year_Last_Improv.
• Year_Record.

The PHT analysis tool is sensitive to these data inputs. Performing analysis using the PHT analysis tool on pavement sections that are missing these data inputs is not recommended.

As part of the PHT analysis, the input data were validated. If input data were missing from the HPMS data, it was possible to supplement the data from other sources if available. Sections that did not contain the required data or other requirements were removed from the analysis.

SUMMARY

This chapter documented the algorithms formulated to perform the RSI application and validation at the project level, the network level for two selected State transportation departments, and the strategic level using the PHT analysis tool to demonstrate the RSI concept.

Chapters 3, 4, and 5 describe how the RSI concept evolved beyond considering the whole life of the pavement system as illustrated by the algorithms documented in this chapter to further considering and determining the optimal decision for the pavement system. Instead of determining the RSI numerics of preservation, rehabilitation, and reconstruction RSI based on the construction thresholds, the RSI numerics would be established by the optimal timing of treatments in order to produce the LLCC for the pavement system. This evolution represents a shift from a change in terminology, as was initially presented, to a change in approach. The details of this evolution are presented in chapters 3, 4, and 5.

¹ Average annual daily traffic.

² Functional system.

³ National Highway System.

⁴ Pavement serviceability rating.