Computational & Technology Resources
an online resource for computational,
engineering & technology publications 

CivilComp Proceedings
ISSN 17593433 CCP: 89
PROCEEDINGS OF THE SIXTH INTERNATIONAL CONFERENCE ON ENGINEERING COMPUTATIONAL TECHNOLOGY Edited by: M. Papadrakakis and B.H.V. Topping
Paper 60
Bridge Failure Analysis Using Fuzzy Fault Tree Methods H.M. AlHumaidi^{1} and F. Hadipriono Tan^{2}
^{1}Department of Civil Engineering, Kuwait University, Kuwait
H.M. AlHumaidi, F. Hadipriono Tan, "Bridge Failure Analysis Using Fuzzy Fault Tree Methods", in M. Papadrakakis, B.H.V. Topping, (Editors), "Proceedings of the Sixth International Conference on Engineering Computational Technology", CivilComp Press, Stirlingshire, UK, Paper 60, 2008. doi:10.4203/ccp.89.60
Keywords: bridge failure, bridge collapse, fault tree analysis, fuzzy fault tree, fuzzy logic, fuzzy set, fuzzy model, enabling causes, triggering causes, procedural causes.
Summary
A new classification of the causes of bridge failure is introduced in this study. The
causes of bridge failures are classified into procedural bridge failure causes,
triggering bridge failure causes and enabling bridge failure causes. Procedural bridge
failure causes are related to management actions and the strategies that impact on
other causes of bridge failure such as triggering bridge failure causes and enabling
bridge failure causes. Triggering bridge failure causes such as wind conditions,
earthquakes and floods are external to the bridge. Enabling bridge failure causes are
internal to the bridge such as foundation related causes, column related causes,
girder related causes and deck related causes.
Probabilistic fault tree analysis is based on the implementation of historical data. Implementation of probabilistic fault tree analysis into bridge failure is not practical due to many reasons. One of the reasons is that bridge failures are unique in their nature, thus implementation of historical data into unprecedented failures of bridges is questionable. Another reason is that management uses linguistic terms to express their opinion in terms of causes of bridge failure and their effectiveness on the likelihood of bridge failure. Quantification of such linguistic terms using probabilistic fault tree analysis is difficult. The fuzzy fault tree analysis is introduced in this study as a method to analyze the likelihood of bridge failure using Baldwin's fuzzy logic model. The alphacut method is implemented to capture the cause of bridge failure and the effectives of these causes on the likelihood of a bridge to fail. The alphacut method is used in Baldwin's rotational fuzzy logic model. Baldwin's rotational fuzzy logic model uses ramp functions where linguistic terms are represented using fuzzy membership values set by Baldwin's model. All linguistic hedges are represented by the powers of the membership functions. The strength of Baldwin's rotational model is that all membership functions overlap, which makes fuzzy operations of the AND and the OR gates easy. Furthermore, the likelihood of bridge failure membership functions can be easily compared to Baldwin's predefined membership functions to assess the severity of the likelihood of bridge failure. The weakness of Baldwin's rotational model lies in the fact that the membership functions of the terms: very negative, negative and fairly negative are fixed and the end user cannot change these membership functions. The likelihood of bridge failure is calculated as is the weighted average of the likelihood of bridge failure. The likelihood of bridge failure curve is compared to the fuzzy models curves of fairly likely bridge failure, likely bridge failure, and very likely bridge failure. Furthermore, the weighted average calculation can provide management with a quantitative measure of the likelihood of bridge failure. A smaller weighted average calculation means a higher likelihood of bridge failure. purchase the fulltext of this paper (price £20)
go to the previous paper 
