COMPUTER-AIDED ANALYSIS OF ACCIDENT TREES

侯福均,肖贵平,李欣

DOI: https://doi.org/10.3969/j.issn.1009-6094.2001.02.007

2001-01-01

Abstract:The accident tree analysis is an important method for system safety engineering and management. It structures the relations between events in a system into a Boolean logic model that leads to accident causation. It is unusually versatile in that it allows dynamic considerations to be considered, sensitivity analysis performed, and the results of analysis quantified. It is well known that the quantitative calculations of accident tree are complex and time consuming. With the increase of the complexity of accident trees, computer-aided analysis of accident trees has become an essential part of the process of system safety analysis. In the cases of quantitative and qualitative calculations of accident tree, finding the minimal cut sets is very important. To date, many analytical programs of accident tree exist. In practice, a complex accident tree always has a large scale and an unpredictable number of minimal cut sets, which makes the traditional algorithms developed with static data structure little applicable. The present paper introduces the implementation of the algorithms of analysis of accident tree based on VC++(MFC)'s expandable Data Structure. The memory structure for the data of accident tree and the flow chart of Fussell algorithm are discussed (Table 1 and Fig.1). The Boolean logic structure of an accident tree is depicted graphically by the tree diagram (e.g. Fig.2). Automatically drawing of tree diagram with computer is another important thing in the process of computer-aided analysis of accident tree. The present paper presents a practical method for automatically drawing of accident trees with computer, which is named brick-stockpiling approach. The components of accident tree and the variable definition of the components by the brick-stockpiling method are discussed (Figs.3 and 4). Taking the advantage of this approach, automatically drawing of accident trees with computer becomes easy. Ideas and algorithms are also given, which will be helpful to those who dedicate to mapping other kind of trees and diagrams. In the end, an example is presented (Table 2 and Fig.5).