Constructing a Fault Tree

In order to properly construct a fault tree, the analyst must first posses extensive knowledge of the system or process under consideration. If such knowledge is lacking, then the process must include in-depth participation from the design community, as well as from other applicable organizational elements within the company (e.g., quality and reliability, operations engineering, facility operations). 

The analyst must obtain a clear understanding of the thought process behind the design of the system, as well as any operational criteria that affect system output. 

The creation of a fault tree begins with the identification of the top event. This event can be as broad and general as total system failure or as narrow and specific as component X malfunction. This top event will be placed at the top of the tree, and all subsequent events that lead to the main event will be placed as branches on the 

3 The House The house configuration is used to identify a normal event that occurs during system operation. It is an event that either occurs or does not occur, such as turning a switch on or off. It should be noted that, if either the ON state or the off state is possible during normal system operation, then the possible effect of both on the top event should be considered. 

4 The Diamond An event in the fault tree that is considered undeveloped is represented by a diamond. Use of this symbol identifies an event that the 

---

A runaway chemical reaction can occur if coolers fail (A) or if there is a bad chemical batch (B). Coolers fail only if both cooler 1 (C) and cooler 2 (D) fail. A bad chemical batch occurs if tliere is a wrong mi. (E) or a process upset (F). A wrong mi. occurs only if tliere is an operator error (G) and instrmiient failure (H). Construct a fault tree and obtain tlie minimum cut sets. 

Fault Tree. A fault tree is a logical model that graphically portrays the combinations of failures that can lead to a particular main failure or accident of interest. A fault tree starts with a top event, which is usually a hazard of some kind. The possibility of the hazard must be foreseen before the fault tree can be constructed. A fault tree helps reveal the possible causes of the hazard, some of which may not have been foreseen.2... 

Wu (2007) has constructed a fault tree for the top event SBLC failure on demand , see Figure 1. 

It is possible to construct a fault tree of such generality, breadth, and controlled redundancy that it can be used effectively—with minor adjustments—in every investigation of actual or potential accidents. The new professional does not need to be trained how to construct fault trees, only how to use one particular, readily available version that has been constructed, reshaped, and checked by literally hundreds of people. Even the least experienced trainee Immediately inherits this compact, pervasive technology. 

In constructing a fault-tree to assist in cause analysis, firstly the event-the accident or failure - must be identified. Secondly, all the proximate causes (contributory factors) must be investigated and identified. Thirdly, each proximate cause (i.e. each branch of a contributory factor) must be traced back to identify and establish all the conceivable ways in which each might have occurred. Each contributing factor or cause thus identified is then studied further to determine how it could possibly have happened, and so on, until the beginning or source of the chain of events has been highlighted for each branch of the fault-tree. 

The Fault Tree Handbook, Chapter VIII, has an excellent example of how to construct a fault tree of a pump-motor pressure system, as shown in Figure 7.5. The handbook describes the problem (Roberts et al., 1981). 

---