Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Computational Science, Engineering & Technology Series
SOFT COMPUTING METHODS FOR CIVIL AND STRUCTURAL ENGINEERING
Edited by: Y. Tsompanakis and B.H.V. Topping
Modelling Life-Cycle Performance of Existing Structures and Infrastructures under Uncertainty: Emphasis on Condition and Safety Profiles
D.M. Frangopol1 and L.C. Neves2
1Department of Civil and Environmental Engineering, Engineering Research Center for Advanced Technology for Large Structural Systems, Lehigh University, Bethlehem PA, United States of America
D.M. Frangopol, L.C. Neves, "Modelling Life-Cycle Performance of Existing Structures and Infrastructures under Uncertainty: Emphasis on Condition and Safety Profiles", in Y. Tsompanakis and B.H.V. Topping, (Editor), "Soft Computing Methods for Civil and Structural Engineering", Saxe-Coburg Publications, Stirlingshire, UK, Chapter 12, pp 313-315, 2011. doi:10.4203/csets.29.12
Keywords: bridge management, Markov process, simulation, maintenance, updating.
The optimization of management strategies for existing structures and infrastructures requires the use of reliable prediction models for future deterioration. Over recent decades, several structural deterioration models have been developed, using both discrete and continuous performance indicators. The uncertainty in civil infrastructure deterioration requires the explicit use of probabilistic models. In this chapter, discrete and continuous models are presented and discussed. The discrete models used are based on transition Markov matrices, calibrated using a maximum likelihood optimization procedure. The second model is based on a multi-linear deterioration function, considering all parameters defining these functions as probabilistic. The relation between these models is evaluated and a methodology to define the probabilistic parameters defining the continuous models using discrete information is presented. Performance is defined in terms of both the condition index, resulting from visual inspections, and the safety index, resulting from structural assessment.
The effects of maintenance actions on life-cycle performance, in terms of both condition and safety are analyzed. Both proactive and reactive maintenance actions are considered. Preventive maintenance actions are applied at regular probabilistic time intervals, reducing deterioration for a period of time. Reactive maintenance actions are applied when performance reaches a predefined threshold, resulting in an improvement in performance. In order to include information collected during the life of structures in the performance predictions, Bayesian updating methods are also discussed, in particular in conjunction with multi-linear probabilistic deterioration models. The proposed model allows the definition of a deterioration profile for a single structure, based on prior information obtained for all structures in a network combined with results of bridge inspections.
The proposed approach is applied to a set of reinforced concrete highway bridges in Portugal, using information collected from visual inspections. The models are developed using a combination of real data with, when necessary, expert judgment. The results presented show the potential of parameter estimation methods defined for discrete models in the characterization of continuous probabilistic models as well as the impact on performance of both proactive and reactive maintenance actions. The possibility of combining the initial probabilistic models with the results of inspection is also discussed.
purchase the full-text of this chapter (price £20)