Fault tree is a powerful tool to model system hazards in a deductive, failure based approach. They can be a qualitative model, that often is evaluated quantitatively. The objective of the fault trees is to pinpoint the events or succession of events, that can lead to the occurrence of a failure, also called, top event.
This is a systematic backward step process, in opposite to FMECA , which is a bottom-up approach.
I got introduce to Fault Trees by the R package tutorial of openreliability.com people. There, they described how to do FTA in the statistical programming language R. The topic is fairly new to me, and I have never used it professionally. I am surprised by all the possible applications that offer. Considering seriously to test it further.
Let’s look at some use cases for the Fault Tree analysis.
1. Understand the logic leading to the top event
Visualize the paths that can contribute to an event, and quickly check if the system is lacking redundancy. In case that we add redundancies check how these are affecting the critical path. It can also highlight paths that are based exclusively on human intervention. Is this the desired design?
2. Prioritize the contributors leading to the top event
In the case that the FTA is quantified, indicate which are the events contributing the most to the top event. It would be desirable to prioritize (cost and resources) areas were the contributions are orders of magnitude superior to others and/or clusters of events relating to the same subsystem.
3. As a proactive tool to prevent the top event
In case to carry on proactive maintenance or upgrades to the system, is possible to choose the area that will need the most attention. Many different maintenance procedures claim a benefit. FTA can quantify objectively this benefit in relation to their top event, hence being able to asses one action over other.
4. Monitor the performance of the system
New information can be added and re-evaluate the FTA. For example new failure data or correction of underlying assumptions. With this is possible to monitor trends and time dependent effects, like aging.
5. As a diagnosis tool to identify and correct causes of the top event.
Let’s assume that we have a good FTA already done and the top event or intermediate steps occur (failures). Afterwards is possible to asses quickly which are the most probable causes for that failure to occur, and which mitigation areas shall be prioritized. Following Point 3 of this guide, is possible to plan some corrective actions and evaluate their contribution to the failure necessary to prevent.
6. Assist in designing a system
When planning a corrective maintenance, FTA can be used to evaluate different design options based on their contribution in relation to the top event. In case the data is not available, it can be generic or “guessed” from past behaviour systems.
One consideration for FTA is, like FMECA, that is a tool that needs to be made prior to the failure in order to be effective. Once a failure occur, rushing an FTA exercise will probably give you sub-optimal results. Therefore sounds like a good idea that the Reliability Engineer or a reliability program in-house, takes into account the identification of system critical events and go through a FTA exercise prior to when the failure occurs.
If there is no plan before the failure, when inevitable occurs, rushing won’t help.
This post is based on the first chapter of the available online “Fault Tree Handbook with Aerospace Applications” published by NASA, as well as, “Practical Reliability Engineering” by Patrick D. T. O’Connor
Do you have experience with Fault Tree Analysis? Do you see more applications than the mention above? Will you recommend FTA?