Fault Tree Analysis Diagrams

This FTA diagram example was redesigned from the Wikipedia Commons file: Example Code Fault Tree.jpg. [commons.wikimedia.org/ wiki/ File:Example_ Code_ Fault_ Tree.jpg]
"Use of Software Fault Tree Analysis.
Any SFTA must be preceded by a hazard analysis of the entire system. The information in the hazard analysis identifies those undesired events in the system that can cause serious consequences. It should be noted that in comple x systems not all hazards can be predetermined.
In this respect the technique does not claim to produce consistent results irrespective of the analyst. It is dependent on the judgment of the individual as to when to stop the process and which hazards to analyze.
The SFTA can be used at different stages of the software life cycle, beginning at the software requirements phase. At that early stage, only the bare minimum of the software will be defined, though the system will usually have a preliminary design. Only a very high-level SFTA will be able to be performed. However, as the software design progresses, and the code is developed, the SFTA can be “fleshed out” with the new information. High-level blocks can be broken down into specific components (modules, classes, methods, etc.)." [NASA Software Safety Guidebook. hq.nasa.gov/ office/ codeq/ doctree/ 871913.pdf]
The FTA diagram example "Code fault tree" was created using the ConceptDraw PRO diagramming and vector drawing software extended with the Fault Tree Analysis Diagrams solution from the Engineering area of ConceptDraw Solution Park.

The FTA diagram sample "Fault tree analysis - Insulin delivery system" was redesigned from the illustration of "CMSI 641: Introduction to Software Engineering. Design of Critical Systems. B.J. Johnson. 2005. Loyola Marymount University".
"Another way of assessing hazards is using fault tree analysis. In this process, each of the identified hazards is covered by a detailed analysis to find out what might cause it. Either inductive or deductive reasoning is applied. In the case of software hazards, the usual focus is to determine faults that will cause the system to fail to deliver a system service, such as a monitoring system. A "fault tree" is constructed to link all the possible situations together, to help identify the interrelationships of the failures, which modules may cause them, and what "trickle-down effects" there might be. Here is an example of a fault tree, as applied to the Insulin delivery system from Sommerville...
Note that this tree is only partially complete, since only the potential software faults are shown on the diagram. The potential failures involving hardware, such as low battery, blood monitor or sensor failure, patient over-exertion or inattention, or medical staff failure are noticeable by their absence.
The fault tree and safety specification processes are two ways of helping with system risk assessment tasks. Once the risks are identified, there are other assessments that need to take place. First, the likelihood of the risk occurrance must be assessed. This is often quantifiable, so numbers may be assigned based on things like MTBF, latency effects, and other known entities. There may be other non-quantifiable contributors to the risk likelihood, however, such that these must be assessed and estimated by experts in the domain. (Don't short-change this process when dealing with critical systems!) Finally, the risk assessment must include the severity of the risk, meaning an estimation of the cost to the project in the event the risk item actually does occur. "Cost to the project" means all associated costs, including schedule delays, human injury, damage to hardware, corruption of data, and so on."
[myweb.lmu.edu/ bjohnson/ cmsi641web/ week15-2.html]
The FTA diagram example "Fault tree analysis - Insulin delivery system" was created using the ConceptDraw PRO diagramming and vector drawing software extended with the Fault Tree Analysis Diagrams solution from the Engineering area of ConceptDraw Solution Park.