Abstract
Fault tree construction is one of the most important techniques in system failure diagnosis. It is also a necessary task for both predicting system reliability & safety and system improvement. The current trend is to develop such construction towards full automation. Hence, the underlying method to carry out such construction is called synthesis(or superimposition)method. It is in fact based on that the fault tree for a system is the synthesis of certain mini-trees of units. Hence, the unit model for each unit will be established first in the way that such unit model will be acceptable in quality by majority of analysts. Currently, many synthesis methods have been proposed. The difference among such methods lies in 1) the differently established unit models, 2) whether certain special subsystems such as control loops, trip loops have been taken as modules in advance, 3) the difference of modularization of such special subsystems in case modularization has been taken into consideration and 4) whether the system has been decomposed into several plant sections of smaller size to facilitate fault tree construction, to compare them and finally to. Thus the quality of the constructed fault tree lies in 1): the quality of unit models, 2) whether the above special subsystems should better be modularized, 3) the quality of modularization and 4) the quality of decomposition in case decomposition is taken. The purpose of this thesis is to survey and study such synthesis methods in the current periodicals and finally proposed proper or better ones(s). The proposed method(s) will be illustrated in terms of three real examples.