Complex Risk Analysis of System-Level Effects
In product development, design FMEAs play an integral
role in identifying the most critical failure modes in a
product and driving design improvements. Furthermore, one
of the most important functions of an FMEA is to assure
that the failure modes with the most severe effects are
addressed by the design team. From a risk analysis perspective,
it would be very beneficial to quantify the probability
of occurrence of those severe effects as a function of the
probability of occurrence of their underlying failure modes
and causes. This can be achieved with the use of fault trees.
However, the task of building a fault tree for a high severity
effect can become challenging when dealing with complex
systems where multiple FMEAs are performed for different
subsystems and components, and the same severe effect can
be present in multiple locations.
In this article we’ll use an example to illustrate how
the Synthesis platform can be used to build a fault tree
of an effect that appears in multiple FMEAs in
If you have Synthesis installed on your computer, you can
download and view the example file (10 MB, *.rsrp)
Consider the case of a turbofan engine whose system hierarchy
was created in Xfmea as shown in the next figure.
FMEAs were performed on many different items within the
hierarchy. Items identified with the blue "F" icon have
a completed FMEA. Overall, there are more than thirty FMEAs
for the system.
The "Uncommanded Engine Shutdown" failure effect appears
multiple times across the different FMEAs, and it is a major
safety concern. The next figure shows an example of this
effect in a single FMEA.
Given the obvious safety concern regarding this effect,
a question that might arise is "What is the probability
of occurrence of this effect across the system?" To answer
this, you can automatically build a fault tree for this
(or any other) effect in BlockSim.
In BlockSim, we choose Insert > Build from
Synthesis > Build Effect FTs from Synthesis. We then
enter "uncommanded" as a criterion to filter the effects
as shown in the next figure.
Note that the results returned include text such as "uncommanded
engine shutdown," "uncommanded engine shut down," "uncommanded
IFSD" ("IFSD" stands for "in-flight shutdown") and "uncommanded
acceleration." This highlights the importance of consistency
when a team is performing an FMEA. In this case, we select
all effects except for the one with the phrase "uncommanded
In the generated fault tree, the top gate represents
the effect of interest (represented by any of the selected
effect descriptions), the next level gates represent the
failure modes associated with the effect (in any of the
FMEAs for the system), and the end events represent all
the causes of the failure modes. The next figure shows part
of the generated fault tree.
Each event in the fault tree contains the cause occurrence
probabilities as defined in Xfmea. As an example,
the next figure shows the block properties window of the
cause "Impending fuel filter bypass switch failure" that
has a 1 in 100,000 probability of occurrence.
Using all that information, we can now estimate the probability
of an uncommanded engine shutdown at 3,000 hours by choosing
Show Results in the control panel, then selecting
Unreliability and entering 3000 hours as shown
in the next figure.
The diagram shows that the probability of an uncommanded
engine shutdown at 3,000 hours is about 0.1319%.
We can now follow the same process in order to create
a fault tree for other high safety risk effects. As an example,
we have created fault trees for effects "damage to a/c,"
where a/c is the aircraft, and "loss of power." Now all
three fault trees can be combined using subdiagrams in order
to obtain the overall probability of occurrence of any of
these critical effects at 3,000 hours, which is found to
be about 0.1768% as shown in the next figure.
In this article we illustrated how to build a fault tree
in BlockSim from an effect that appears in multiple
FMEAs in Xfmea. Then we calculated the probability
of occurrence of that effect by using the probability of
occurrence of each underlying cause. This process can be
used to build a fault tree of all high severity effects
in a system and the created fault trees can be combined
using subdiagrams in order to calculate the overall probability
of occurrence of a high severity effect.