Simplified Techniques for Evaluation and Interpretation of Pavement Deflections for Network-Level Analysis

CHAPTER 3 - Assessment of Data

REVIEW AND ASSESSMENT OF DATA FROM LTPP STUDIES

Data Selection Criteria

The goal of this task was to review data from LTPP studies that are relevant to the scope of this study. Of particular interest was the assessment of LTPP sites that had sufficient deflection information, site characteristics (structure, climate conditions, traffic, and age), and historical performance data consisting of load-related distresses and roughness levels. Data from both General Pavement Studies (GPS) and Specific Pavement Studies (SPS) were considered. SPS sites are particularly valuable due to the extensive material testing and performance data concurrently available. In addition, sites from the Seasonal Monitoring Program (SMP) provided useful information for studying environmental effects, as well as the timing and frequency of deflection testing for network-level coverage.

One objective of a PMS is to establish long-term maintenance and rehabilitation (M&R) strategies for the network. Therefore, only those pavement sections for which the database included at least 5 years' worth of data were included in the analysis database.

Data Elements Needed for Analyses

The following data elements were identified as necessary for the analysis:

• Performance data:
  • AC surfaces.
  • Rutting.
  • International Roughness Index (IRI).
  • Fatigue cracking.
  • Transverse cracking.
  • Longitudinal cracking.
  • Alligator cracking.
  • PCC surfaces.
  • IRI.
  • PCC faulting at joints.
  • PCC faulting at cracks.
  • Transverse cracking.
  • Longitudinal cracking.
  • Durability cracking.
  • Number of corner breaks.
• Pavement structure:
  • Pavement layer type.
  • Representative thickness for a layer in a section.
  • Type of material (AC, PCC, etc.).
• Pavement deflection data:
  • Peak deflection values.
  • Drop height.
  • Peak load (peak load equals peak plate pressure).
  • Lane number.
  • Lane location.
  • Measured pavement surface temperature.
  • Time of day.
• Climatic data:
  • Average temperature.
  • Average precipitation.

Table 3 provides a summary of data types that were assessed in this investigation, along with the corresponding LTPP data tables containing the required data elements.

Table 3. LTPP database tables used to support data collection.

Analysis Database Development

The LTPP Standard Data Release (SDR) 23 database (the most up-to-date version at the time this study was conducted) was examined for the identified data elements to determine the extent of data availability.⁽¹⁾ Data were extracted from the LTPP database, imported into a Microsoft Access^® analysis database, and stored in various data tables. These tables were set up according to the principles of the relational databases so that they could be manipulated and linked together for various analysis purposes.

Data Review Process

Data for the sites that satisfied the data selection criteria were reviewed for quality, reasonableness, and availability in light of supporting the evaluation of identified deflection techniques.

The first step in the data review process was to eliminate test sections that had less than 5 years of performance and/or deflection data from the beginning of the monitoring period. In the sequence, quality and reasonableness were evaluated for all sections selected. The performance plots and deflections versus time were then visually inspected. Outliers and unreasonable trends were identified for a more detailed evaluation. A summary of the data assessment results is presented in the following sections.

Outliers identified in performance trends were marked for possible exclusion from the datasets. Ultimately, whether an outlier was excluded depended on its influence on the performance trend over the years that data were collected and its impact on predicting future performance. Details of this approach are provided in chapter 4 of this report.

Assessment of Data to Support Evaluation of Identified Techniques

All deflection data collected during the experiment were obtained from the LTPP SDR 23 database and assessed.⁽¹⁾ In total, there were 1,027 sections in the SPS and GPS experiments with 5 or more years of performance and deflection data available. These sections include all pavement types - flexible, rigid, and composite. These sections may or may not have been rehabilitated during the course of data collection. Table 4 lists the experiments, number of sections available with deflection and performance data, and average period of time of data coverage.

Table 4. LTPP data availability to support evaluation of identified techniques.

Impact of M&R on Pavement Performance Data

In addition to verifying performance requirements and deflection availability, it was also important to consider any M&R work on the LTPP lane of the analysis sections. When one section is rehabilitated, performance invariably improves, and this affects the outcome of the analysis. Of the 1,027 sections selected, 47 of the flexible sections, 105 of the rigid sections, and 97 of the composite sections had been rehabilitated (almost 25 percent of all sections). An approach to identify these sections and split them based on the rehabilitation work performed (i.e., change in structure and reduction of distress level) was developed. The main reason for performing this type of analysis was the possibility of increasing the number of available representative pavement sections, especially for the more sparsely populated rigid and composite pavement types.

The multiple section analysis was designed to enhance the chances of using sections with rehabilitation work performed during the monitoring period. Usually, any rehabilitation work done on the pavement is indicated in the LTPP database by a change in construction number. This is an event number used to relate changes in pavement structure with other time-dependent data elements. This field is set to "1" when a test section is accepted into the LTPP program and is incremented with each M&R activity.

The first step was to plot the performance data as a function of time for all sections. A simple macro-based code was written to expedite this process. The analysis of performance change is required because sometimes the M&R work is done outside the LTPP lane (e.g., at the shoulder) with no implication on LTPP lane performance.

For this analysis, the IRI was the performance measure used to determine the cutoff date to split the section based on recorded M&R activities. IRI is sensitive to M&R work performed on the pavement, and IRI measurements are more frequent in the LTPP distress survey campaign, which enhances the chances of finding sections that can be successfully split and still provide enough data for analysis.

Figure 20 shows section 06-0509, which illustrates how the section split was done based on rehabilitation work executed on the pavement. This example is a flexible pavement section located on I-40 in California, and the figure shows the variation of roughness over time. There were two rehabilitation interventions executed during the time the section was monitored. Three distinct pavement structures are identified. The first subsection (A) corresponds to the pavement from the original construction. The first split creates another subsection (B) with a new structure due to partial milling of the existing pavement and placing a hot mix recycled asphalt concrete overlay. The second split creates the third pavement subsection (C) due to a single layer of surface treatment applied to the existing surface. These two splits yield the following subsections and pavement structures:

• Structure A has a construction number index of 1, corresponding to the following original pavement structure:
  • Untreated subgrade.
  • Unbound (granular) subbase.
  • Bound (treated) base.
  • HMA concrete.
• Structure B has a construction number index of 2, corresponding to partial milling of the existing surface layer in A, followed by filling and overlay using a hot recycled asphalt concrete mix.
• Structure C has a construction number index of 3, which corresponds to another HMA concrete overlay placed on top of B.

^{1 inch = 25.4 mm
1 mi = 1.61 km}

Figure 20. Graph. Variation of roughness in section 06-0509

After the multiple section analysis was concluded, each of the three new subsections was reevaluated based on the 5-year performance criterion. The first subsection (A) was excluded from the analysis because it had less than 5 years of performance data. Subsections B and C both had more than 5 years of performance data and were therefore included in the project's dataset.

This process was carried out on all LTPP sections selected to support the evaluation of simplified deflection-based analysis techniques. Table 5 shows the final data assessment summary after performing the multiple section analysis. Notice that if the sections that received one or more M&R procedures after being assigned to the LTPP database were excluded altogether from the dataset, the total number of available sections would have been 827. After splitting the sections and including the subsections with 5 years or more of performance data, the total number of available sections increased to 1,033. Accordingly, the multisection analysis minimized the loss of LTPP data points available for the study.

Table 5. Data assessment summary after multiple section analysis.

Analysis of Deflection Data

Deflection data were analyzed for quality and reasonableness. Prior LTPP studies investigated the quality of deflection data, and the recommendations and final tables originating from these studies were used as the source of data mining for this project.⁽²³⁾ After the data were extracted into spreadsheets, each deflection-based technique or parameter was computed. Table 1 lists the deflection techniques investigated in this project.

ASSESSMENT OF DATA TO SUPPORT OPTIMUM TEST SPACING AND FREQUENCY

LTPP Data to Support Optimum Test Spacing Analysis

As part of phase I, the applicability of LTPP data for this type of analysis was verified. One of the concerns with LTPP data was the limited length of the LTPP sections (generally 500 ft (152.5 m)), which is not ideal for a statistical analysis of optimum test point spacing on the network level. A theoretical statistical study was conducted to evaluate the implications of different sampling strategies (i.e., different test spacings) on the expected error of average deflections of homogenous road segments, when compared with comprehensive test spacings. The comprehensive test spacings were defined as the typical 0.1-mi (0.016-km) spacing commonly used by transportation departments for project-level FWD testing. The theoretical statistical study was then compared to field data obtained from State transportation departments.

Supplemental Data to Support Optimum Test Spacing Analysis

Supplemental data were required to investigate optimum FWD test spacings for network-level deflection data collection. FWD data from three State transportation departments were obtained and used to test results obtained from the theoretical statistical study.

LTPP Data to Support Optimum Test Frequency Analysis

Selected sections from the GPS and SPS studies were monitored for temperature and moisture at higher-than-normal intervals for distress, deflection, and longitudinal profile (IRI). These sections were part of the SMP database and were classified as either flexible or rigid pavements. Table 6 shows the classification of these SMP sections and the availability of FWD data from these sections.

Table 6. FWD data availability for SMP sites.

The SMP data sections were scrutinized in the same manner as was used for the GPS and SPS sections for data quality and reasonableness. The objective of incorporating the SMP sections was to provide data elements for the analysis of frequency of deflection measurements for network-level applications. Moreover, the data were used to evaluate and develop guidelines for the optimum time of day and season of year for FWD (or equivalent deflection testing device) data collection.

After initial inspection of data availability, it was found that monthly collection of data in most SMP sections occurred for about 2 years (or a maximum of 3 years). This volume of data reduced the chances for a robust assessment of the optimum frequency of FWD data collection for network-level applications. Therefore, it was decided to complement the dataset with additional LTPP sections from the dataset defined for the study of deflection-based techniques. The approach for using both datasets is further described in chapter 5 of this report.

LTPP Data to Support Time of Day and Day of Year Analysis

Different agency practices and provisions related to time-of-day and time-of-year data collection and analysis were reviewed. For example, the UK highway agency has a requirement for network-level deflectograph data collection from March 15 to June 15 and from September 15 to November 30. This requirement is primarily based on the climate in the United Kingdom. For the United States, different climate zones will most likely have different periods during which data collection will be most favorable to capture useful network-level deflection data. The subset of the LTPP database used to evaluate the deflection techniques was also used to analyze FWD data for different climatic zones that were taken during different months to determine favorable data collection periods for different regions.