MARKOV DECISION PROCESS-BASED PREDICTION OF IN-SERVICE RESIDUAL ULTIMATE STRENGTH CAPACITY OF A REINFORCED CONCRETE BRIDGE DECK
The discrete-state, discrete-time Markov Decision Process (MDP) has been applied for the prediction and assessment of the future in-service residual strength capacity of a standard reinforced concrete bridge deck subjected to dominant deterioration processes. This stochastic model predicts the future condition states of the bridge elements based on their current condition states through a multiplicity of discrete-state, discrete-time conditional, singly-stochastic transitional probability mechanism. There are four specific environments (i.e. benign, low, moderate and severe) into which the bridge elements have been divided to allow for differences in deterioration rates due to the differences in their degrees of severity. These are occasioned by the identified dominant operating practices such as age of deck,average daily truck traffic, chloride applications, climatic factors and construction procedure which are the drivers of the deterioration processes that have also been modeled and calibrated through the derived Markov chains to produce corresponding non-dimensional deterioration indices for the design life of the bridge or the period for research consideration. The Markov chains which are singly stochastic take the Jordan canonical matrix form. The differences in deterioration rates have been captured by introducing multiple Markov chains. These deterioration indices are applied to obtain the actual depth of deterioration and the normalized remaining ultimate strength capacity and the corresponding actual ultimate strength capacity of each of the bridge elements. Since bridges residual capacity is non-steady, the results enable bridge evaluators and decision makers to determine when to close the bridge to traffic for reconstruction or major maintenance. The output also provides basis for the implementation of a comprehensive bridge management system that is analytic and non-empirical.
Markov Chain Stochastic Deterioration Bridge Deck Residual Capacity Probability Matrix
ADEYEMO, Emmanuel A. AKEJU, Timothy A. I
Pheman Peniel Consultants Lagos, Nigeria University of Lagos Lagos, Nigeria
国际会议
上海
英文
229-238
2007-05-16（万方平台首次上网日期，不代表论文的发表时间）