Operations research and data mining applications in pavement management

dc.contributor.advisor Smadi, Omar
dc.contributor.advisor Dong, Jing
dc.contributor.advisor Wood, Jonathan
dc.contributor.advisor Madson, Katherine
dc.contributor.advisor Dixon, Philip
dc.contributor.author Abukhalil, Yazan
dc.contributor.department Department of Civil, Construction and Environmental Engineering
dc.date.accessioned 2022-11-08T23:56:23Z
dc.date.available 2022-11-08T23:56:23Z
dc.date.issued 2021-08
dc.date.updated 2022-11-08T23:56:23Z
dc.description.abstract Transportation agencies manage and operate large pavement networks with the goal of maintaining high performance levels through cost-effective decision-making. To be able to achieve this goal, transportation agencies use decision support tools (DSTs) to evaluate networks’ current condition, predict future condition, and find optimum maintenance and rehabilitation (M&R) budget allocation strategies. DSTs have drawbacks including subjectivity and inefficiency. Given the drawbacks of the currently used DSTs, and since transportation agencies started to move toward more efficient and objective data-driven decision-making approaches with the continuous pavement network data collection, this dissertation aims to develop data-driven DSTs for pavement management systems (PMS) that transportation agencies can use to improve the efficiency of their decision-making process. To help achieve this goal, three objectives are addressed in this dissertation. A bootstrap-based binary integer programming algorithm is designed to allocate the budget over an extensive pavement network under budget and performance constraints. This algorithm succeeded in significantly reducing the computational cost required for budget allocation in PMS. Furthermore, data-driven decision trees (DTs) are developed for M&R selection in PMS using the Classification and Regression Trees algorithm. These trees were able to capture patterns from historical M&R actions with relatively high accuracy levels. The last component of this dissertation addresses a data quality issue, the underreporting of M&R projects. A logistic regression-based approach for detecting unreported pavement treatments from historical condition trends is suggested. High validation accuracies were achieved for the three logistic models developed for the interstate, U.S., and state route systems in Iowa.
dc.format.mimetype PDF
dc.identifier.doi https://doi.org/10.31274/td-20240329-563
dc.identifier.uri https://dr.lib.iastate.edu/handle/20.500.12876/2vaZ4Ygr
dc.language.iso en
dc.language.rfc3066 en
dc.subject.disciplines Civil engineering en_US
dc.subject.disciplines Transportation en_US
dc.subject.disciplines Management en_US
dc.subject.keywords Budget allocation en_US
dc.subject.keywords data mining en_US
dc.subject.keywords machine learning en_US
dc.subject.keywords optimization en_US
dc.subject.keywords pavement management en_US
dc.title Operations research and data mining applications in pavement management
dc.type dissertation en_US
dc.type.genre dissertation en_US
dspace.entity.type Publication
relation.isOrgUnitOfPublication 933e9c94-323c-4da9-9e8e-861692825f91
thesis.degree.discipline Civil engineering en_US
thesis.degree.discipline Transportation en_US
thesis.degree.discipline Management en_US
thesis.degree.grantor Iowa State University en_US
thesis.degree.level dissertation $
thesis.degree.name Doctor of Philosophy en_US
File
Original bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
Abukhalil_iastate_0097E_19733.pdf
Size:
2.58 MB
Format:
Adobe Portable Document Format
Description:
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
0 B
Format:
Item-specific license agreed upon to submission
Description: