Date of Award
Doctor of Philosophy
Civil, Construction, and Environmental Engineering
Civil Engineering (Intelligent Infrastructure Engineering)
Pavement management, planning ahead for the maintenance and repair of the roadway, is dynamically developing with evolving technology. To collect reliable, accurate, and complete data, agencies must perform quality management (QM) on their pavement data. QM is defined as the process to achieve high-quality output. It is a procedure to ensure reliability, accuracy, and completeness of data. Pavement management engineers have been performing QM on pavement condition data for decades. To maintain a certain level of pavement quality, the Federal Highway Administration (FHWA) has asked state departments of transportation (DOTs) to report all the data processes they follow in their pavement management practices in a document called a data quality management plan (DQMP). A DQMP must contain information regarding the following sections: (1) data collection equipment calibration and certification, (2) certification process for persons performing manual data collection, (3) data quality control measures, (4) data sampling, review and checking processes, and (5) error resolution procedures and data acceptance criteria.
Sampling plays a significant role in QM and it is one of the main parts of DQMP. It is necessary to select the appropriate and representative sample size to ensure that the budget and resources are not being excessively used for large sample sizes, and also the samples represent the population to be able to generalize from their results. Currently, in pavement management, QM sample size is selected based on experience, and using rules of thumb usually 5% or 10% of the population while sample size calculation should be independent of population size.
This dissertation aims to improve the QM of pavement management data by reviewing multiple DOTs DQMPs and detecting the gaps and needs in the state of practice, and also by implementing the power analysis method to select a representative sample. For the first objective of this research, a survey was sent out to DOT pavement management engineers requesting information about their pavement data QM processes. From the 31 states that responded, 28 DQPMs were collected and studied. The best practices based on coverage of the FHWA requirements, the depth of discussion in each DQMP section, and how the information can contribute to pavement QM were identified, and the needs to improve overall QM procedures and ways to enhance DQMPs were explained. Few states discussed sampling in their DQMP which makes the necessity of the second objective of this dissertation more apparent
The second part of the dissertation uses a binary pass/fail quality assurance (QA) data. The QA was done on pavement condition data. Priori power analysis was done using the three main factors of sample size calculation (1) the effect size, which is calculated by the difference of the means of the experimental and control conditions and the variability of this difference across subjects; (2) significance level, which is the confidence or risk level based on the Central Limit Theorem; and (3) the statistical power, which is the probability of rejecting the null hypothesis when it is false.
The final part of the dissertation validates the results of priori analysis using a Monte Carlo simulation (MCS) on populations with different characteristics and various sample sizes.
In conclusion, priori analysis can be used for calculating sample size. The results of this study can be used not only for pavement management but also for other fields as well; from bridge, engineering to water resources, and environmental tests.
Bazargani, Bahareh, "Improving quality management of pavement condition data" (2020). Graduate Theses and Dissertations. 18095.
Available for download on Saturday, August 28, 2021