Reliability-based fatigue assessment of existing steel bridges
Time: Fri 2020-11-13 13.00
Location: Videomöte med Zoom - https://kth-se.zoom.us/j/69922859300, Du som saknar dator/datorvana kan kontakta Johan Leander firstname.lastname@example.org / Use the e-mail address if you need technical assistance, Stockholm (English)
Subject area: Civil and Architectural Engineering, Structural Engineering and Bridges
Doctoral student: Ruoqi Wang , Bro- och stålbyggnad
Opponent: Professor John Dalsgaard Sørensen, Aalborg University, Denmark
Supervisor: Associate Professor John Leander, Byggteknik och design; Professor Raid Karoumi, Bro- och stålbyggnad
Fatigue is among the most critical forms of deterioration damage that occurs tosteel bridges. It causes a decline of the safety level of bridges over time. Therefore,the performance of steel bridges, which may be seriously affected by fatigue, shouldbe assessed and predicted. There are several levels of uncertainty involved in thecrack initiation and propagation process; therefore the probabilistic methods canprovide a better estimation of fatigue lives than deterministic methods. Whenthere are recurring similar details which may have correlation with each other andbe regarded as a system, there are distinct advantages to analyze them from asystem reliability perspective. It allows the engineer to identify the importance ofan individual detail or the interaction between details with respect to the overallperformance of the system.
The main aim of this licentiate thesis is to evaluate probabilistic methods for reliabilityassessment of steel bridges, from both a single detail level and a systemlevel. For single details, an efficient simulation technique is desired. The widelyapplied Monte Carlo simulation method provides accurate estimation, however, isvery time-consuming. The Subset simulation method is investigated as an alternativeand it shows great feasibility in dealing with a multi-dimensional limit statefunction and nonlinear crack propagation. For larger systems, the spatial correlationis considered between details. An equicorrelation-based modelling approachhas been proposed as supplement to common simulation techniques to estimate thesystem reliability analytically and significantly reduce the simulation time. Withcorrelation considered, the information of one accessible detail could be used topredict the status of the system.
While reliability analysis aims for a specific safety level, risk analysis aims to findthe most optimal solution. With consequences considered, a risk-based decisionsupport framework is formulated for the selected system, which is presented asa decision tree. It reveals that the decisions based on reliability assessment canbe different from those based on risk analysis, since they have different objectivecriteria.