Study of long-term pavement performance (LTPP): pavement deflections.

This report presents the results of a study of pavement deflections. The study covered all level E falling weight deflectometer (FWD) deflections and associated data in LTPP’s database from Data Release 9.0, November 23, 1998. Although the limited amount of data from unbound material testing was also provided, these data were not screened due to the large variations in the recorded deflections in comparison with bound layer tests.\nThe report covers the screening techniques developed and used to identify data errors and anomalies in the FWD load-deflection database, along with a description of each category of data errors identified. Contrary to prior expectations, the vast majority of these data errors were related to manually input data elements, not the deflections themselves. Approximately 8 percent of the 4.4 million lines, or records, in the pre-autumn 1998 load-deflection database were affected by manual input data errors alone, while less than 0.2 percent appear to be affected by actual load-deflection data anomalies generated by the FWD. Out of the approximately 8 percent of manual input data errors found, around 7 percent were associated with nonprotocol and unreported placement of the deflection sensors along the FWD’s raise-lower bar. Other types of manual data entry errors, each occurring at a rate of less than 1 percent, included incorrect lane designation, station number, date- or time-stamp, test site, drop height, and configuration of the sensors for joint testing on portland cement concrete (PCC) pavements. Deflection reading data errors included deflection basin anomalies and sensor malfunctioning errors. A universally applicable deflection basin screening tool called SLIC was also developed for use on select FWD data file formats.\nThe overall quality of the pre-autumn 1998 FWD database can be characterized as good to excellent.\n