Pavement management is a process that involves many models, frameworks, and other features to be properly implemented. One common requirement across all pavement management systems is the availability of data: having good quality data can result in much more effective management decisions than when having poor quality data. One challenge in assessing data quality is incomplete maintenance and rehabilitation (M&R) records, which means improvements in pavement condition, as shown by condition data, are not always associated with M&R being performed on that pavement. Commensurately, this paper presents an approach and two algorithms for differentiating between improvements in pavement condition because of M&R and those improvements that are a result of measurement variability. The algorithms were trained using data from the Long-Term Pavement Performance database, and an application of the algorithms was demonstrated using data from a U.S. state department of transportation. The results show that the trained algorithms can effectively differentiate between M&R and measurement variability, even when the M&R type is preservation. Additionally, the algorithms showed multiple cases where the recorded year that M&R was complete did not correspond to the year that paving was completed.
James Bryce (Tue,) studied this question.