Nonnormality in the form of skewness and kurtosis was examined in lot acceptance quality characteristics data from seven state highway agencies for their highway construction quality assurance programs. Lot skewness and kurtosis varied significantly. For most lot data sets, skewness values varied in the range of 0.0 ± 1.0, whereas most kurtosis values varied in the range of 0.0 ± 2.0. The analysis also reveals that, on average, 50% of lot test data sets were nonnormal with 15% of lot data sets having skewness greater than ±1.0 and kurtosis greater than ±2.0. This is a significant finding because most state transportation agencies' pay factor algorithms assume normally distributed lot. Further investigation showed that high skewness and kurtosis were associated with higher lot variability. This variability produced misleading results in regard to inflated Type I error and low power for the F-test. However, the t-test was found to be quite robust for distinguishing mean differences. Significant deviation was observed in lot pay factors based on percent within limits between assumed normal data and normalized data. Effects of nonnormal distribution on the lot pay factor were found to be varied on the basis of the specification limits, the distribution of defective materials on the tails in the case of two-sided limits, and the orientation of the nonnormal distribution itself.

<p>  </p> <p>For decades, agencies have collected infrastructure condition assessment data using dedicated equipment that require substantial capital investments and staff time to operate/drive. However, these techniques are challenging to scale network wide. The United States has over 8 million lane miles of roadways which generate almost 3 trillion vehicle miles annually.  Connected vehicles can now provide real-time data on a wide range of parameters such as vehicle speed, location, lane markings, and 3 axis acceleration. This dissertation develops techniques to validate, utilize and leverage connected vehicle data for infrastructure assessment and monitoring.  </p> <p><br></p> <p>Opportunities to employ connected vehicle data were examined in the following areas: quality of lane marking edge lines, width of lanes (particularly temporary lanes in construction zones), and pavement roughness. Quality of lane markings was evaluated using embedded lane keep assist data and equipment. In 2020 and 2021 over 5000 miles of pavement markings were evaluated on Indiana interstates. Comparisons between 2020 and 2021 data showed detection increase from 80.2% to 92.3%.  Although there are no industry standards for lane keep assist data, this study demonstrated both the importance and utility of partnering with the automotive industry to develop shared vision on acceptable lane quality. </p> <p><br></p> <p>A follow-up quantitative study was performed using a LiDAR vehicle to compare LiDAR values with those that are obtained from traditional retroreflectivity measurements used for contract acceptance and maintenance decisions. A comparison of LiDAR intensity to retroreflectivity (the industry standard) on 70 miles of US-52 and US-41 in Indiana was assessed and a linear regression found that the intensity values are comparable to retroreflectivity readings with an R2 of 0.87 and 0.63 for right edge and center skip lines respectively. These results suggest that LiDAR is a viable tool for monitoring of retroreflectivity of pavement markings that are strongly correlated with existing standards, but scale much better than traditional retroreflectivity measurement techniques.</p> <p><br></p> <p>The LiDAR data also provided the opportunity to evaluate how well modern vehicles measure lane width. This dissertation reports on over 200 miles of roadway and when compared to LiDAR and field measurements had a root mean square error of 0.24 feet. This data is valuable for agencies to quickly identify system wide where lane widths fall below acceptable design standards, typically 11-feet. </p> <p> </p> <p>The final connected vehicle data set evaluated was pavement roughness and compared with traditional dedicated vehicles collecting international roughness index (IRI) data. The study evaluated a 20-mile segment in 2022, and showed a linear regression between these data sets had an R2 of over 0.7, suggesting that connected vehicle roughness data can be utilized for network level monitoring of pavement quality. Scalability of these techniques is also illustrated with graphics characterizing IRI values obtained from almost 6 million records to evaluate improvements in Indiana construction zones and over 5,800 miles of I-80 in April of 2022 and October 2022.</p> <p><br></p> <p>Although connected vehicle data for infrastructure assessment is still in its infancy, these case studies demonstrate significant opportunities for public agencies to collect selected system wide infrastructure condition in near real-time, and in many cases at a lower cost than traditional techniques. </p>