Evaluating the Repeatability and Inter-Device Reproducibility of International Roughness Index Measurements

The use of network survey vehicles (NSVs) for the functional evaluation of highway assets has become industry practice around the world. However, owing to the lack of adequate quality control in pavement condition assessment at the network level, substantial variability in roughness measurement can be observed attributed to interchangeable usage of multiple operators, varying segment lengths, presence of local anomalies, and so forth. This study assesses the repeatability and reproducibility of international roughness index (IRI) measurements computed for a wide range of segment lengths and IRI run averages across multiple NSVs. Fifteen runs of IRI measurements were conducted using three NSVs along a six km section of National Highway 44. The results showed that the variability in IRI measurements reduced as the number of runs and the segment length under consideration were increased. Consequently, there exist a combination of measurement mitigation strategies that can be considered by agencies to meet quality assurance standards. The results also demonstrated variability in the repeatability results across NSVs as well as lack of agreement in IRI measurements for different segment lengths. The differential impact of the data collection parameters on IRI measurement is highlighted through a heteroscedastic linear regression framework. Additional analyses explored the impact of following marked paths on run-to-run variability, as well as the impact of structural joints on IRI estimates. Some strategies for determining critical limits for repeatability and reproducibility are also discussed.