Distress Data Consolidation Final Report

Distress data have been collected for more than 10 years through the LTPP project. Sufficient data now are available to begin examining trends in pavement performance. Prior to any rigorous analysis of the LTPP data, however, there is a need to assess the quality of the data. Therefore, the primary objective of this effort was to produce a comprehensive consolidated distress data set to reconcile differences between data collected using different methodologies. Although as many as 15 distinct distresses are included in the DIM for each pavement type, the analysis reported herein focused on the most commonly occurring distresses shown in table 24.

Table 24. Most commonly occuring distresses for various pavement types.

CONCLUSIONS

The following conclusions are noteworthy:

• Based on a thorough review of the LTPP database (release 8.6, October 1998), the authors concluded that two-thirds of the condition survey data are in good shape (i.e., are suitable for inclusion in the consolidated data set and hence, more rigorous analysis).
  
  
• The remaining one-third of the condition survey data was problematic in that there were discrepancies within and between data sets. These survey data will require reevaluation and correction prior to inclusion in the consolidated data set.
  
  
• These discrepancies were categorized into five groups, as shown in table 25.
  
  

Table 25. Categorization of discrepant surveys.

• Discrepant survey data categorized as human error were both quantitative and qualitative in nature. Computational errors occurred in compiling numerical data such as total number of cracks or area of patches. Due to the subjective nature of evaluating visual distress, many surveys were labeled discrepant because of problems in differentiating between the following:
  • Fatigue cracking and longitudinal cracking in the wheel path.
    
    
  • Wheel path longitudinal cracking and non-wheel path longitudinal cracking.
    
    
  • Block cracking and longitudinal and transverse cracking.
    
    
• Although some discrepant survey data were attributed initially to seasonal effects, that classification could not be confirmed as part of this review process.
  
  
• Overall, the condition survey results were found to be independent of data collection methodology: manual and automated (PASCO with PADIAS 1.x and 4.2 software) distress surveys yielded similar results. However, it should be noted that the PADIAS 1.x software used to reduce the pre-1993 distress data did not distinguish between non-wheel path and wheel path longitudinal cracking. Unless these pre-1993 data are re-assessed to determine the quantity of non-wheel path versus wheel path longitudinal cracking, there is some error in the database.
  
  
• Evaluation strategies or erroneous assumptions made at the onset of this analysis accounted for approximately 36 percent of the discrepant data. These included the following:
  
  
  • Statistical analyses that did not account for the baseline measurement of distress in establishing variability measurements (i.e., standard deviation and variance).
    
    
  • Statistical analyses that did not account for nonlinear increase in distress with time.
    
    
  • Identification of maintenance and rehabilitation activities.
    
    
• Approximately 41 percent of the questionable survey data remain unclassified in terms of problem source.
  
  
• Some subjectivity can be eliminated in the preliminary data QC process by automating the process.
  
  

RECOMMENDATIONS

In view of the preceding conclusions, the authors offer the following recommendations:

• Three data tables, one each for HMA, JC, and CRC pavements, should be added to the LTPP database. These tables would house the data that pass the QC process.
  
  
• FHWA is encouraged to use the QC software to expedite the preliminary data analysis process.
  
  
• The nearly one-third of the distress data that were deemed discrepant should be reevaluated using the maps created in the survey process.