Fatigue Test and Prognosis Study of Welded Tubular Joints in Signal Support Structures
| dc.contributor.advisor | Zhang, Yunfeng | en_US |
| dc.contributor.author | Li, Zhen | en_US |
| dc.contributor.department | Civil Engineering | en_US |
| dc.contributor.publisher | Digital Repository at the University of Maryland | en_US |
| dc.contributor.publisher | University of Maryland (College Park, Md.) | en_US |
| dc.date.accessioned | 2013-07-04T05:31:26Z | |
| dc.date.available | 2013-07-04T05:31:26Z | |
| dc.date.issued | 2013 | en_US |
| dc.description.abstract | Steel welded tubular joints have been widely used in traffic signal support structures for economic and aesthetic reasons. However, they are susceptible to fatigue cracking which may lead to structural failure such as collapse, thus pose a threat to driver's safety. To address this issue, this study is focused on fatigue test, modeling and prognosis of the fatigue crack growth in full-scale welded tubular joints of traffic signal support structures. Fatigue test of six full-scale welded tubular joint specimens fabricated based on real design for signal support structure is conducted to obtain crack growth data. Details of test setup and results are reported in this dissertation. Two types of fatigue crack growth models are proposed for two regimes of fatigue crack development in welded tubular joints: the linear elastic fracture mechanics (LEFM) model for the slow crack growth regime (denoted as Stage II here) and the empirical failure model for the rapid crack growth regime (denoted as Stage III). Details of these two models including their mathematical expressions, stochastic parameters, sensitivity analysis and model application, are given in the dissertation. A sensor-driven structural health prognosis procedure that has an explicit stochastic measurement error term and thus can model the sensor performance degradation over monitoring period is proposed. The prognosis procedure involves the Bayesian theorem and Markov Chain Monte Carlo (MCMC) sampling for updating the structural degradation model using sensor data. An extreme value theory (EVT) based tail fitting method is proposed to reduce the heavy burden on data transmission and computing involved in sensor driven prognosis. This method employs moment estimator to calculate the small quantiles of the prognosis results by using a small portion of available sensor data. Finally, fatigue test data acquired in this study are used to examine the proposed fatigue life prognosis procedure. Both the LEFM based fatigue crack growth model and the empirical failure model are studied for fatigue life prognosis application. Prognosis results show that the prognosis procedure is able to provide good estimate of the fatigue crack growth curve of welded tubular joints in signal support structures if certain conditions are met. | en_US |
| dc.identifier.uri | http://hdl.handle.net/1903/14269 | |
| dc.subject.pqcontrolled | Civil engineering | en_US |
| dc.subject.pquncontrolled | Crack growth modeling | en_US |
| dc.subject.pquncontrolled | Extreme value theory | en_US |
| dc.subject.pquncontrolled | Fatigue test | en_US |
| dc.subject.pquncontrolled | Prognosis | en_US |
| dc.subject.pquncontrolled | Sensor degradation | en_US |
| dc.subject.pquncontrolled | Tubular Joints | en_US |
| dc.title | Fatigue Test and Prognosis Study of Welded Tubular Joints in Signal Support Structures | en_US |
| dc.type | Dissertation | en_US |
Files
Original bundle
1 - 1 of 1