AFD Facts
Failure Analysis -- A systematic procedure for identifying the root causes of a failure or
other undesired phenomenon in a system, and for making corrections in a timely
manner.
Failure Prediction -- A systematic procedure for identifying beforehand, and then
preventing, all dangerous or harmful events that might be associated with a system.
Systems in which failures have occurred -- or might occur -- are zones of "poor information." The
reason? Little information is published about negative effects with unknown causes, or about the
causes of dangerous or harmful failures. In fact, such information is often intentionally
concealed.
Without adequate information, it is very difficult to identify the root causes (existing or
possible) of a failure. One must rely on guesswork -- as is the case with traditional failure
methods.
AFD overcomes this obstacle with a core 3-step model, providing unprecedented effectiveness:
For Failure Analysis: Instead of asking "Why did the failure happen?" ask instead: "How can I
make it happen?"
For Failure Prediction: Instead of asking "What failures might happen?" ask instead: "How can I
make all possible dangerous or harmful failures happen?"
Now we can employ a wealth of available information based on what inventors have profited
from since the dawn of mankind: how to make something happen. In other words, we have
converted a failure problem into an inventive problem.
STEP 2: IDENTIFY FAILURE HYPOTHESES. Find a method by which the known or potential
failures can be intentionally produced.
STEP 3: UTILIZE RESOURCES. Determine if all the components necessary to realize each
hypothesis are available in your system, or if they can be derived from what is available:
. . . and more
The principle difference between AFD and conventional techniques, such as Failure Mode and
Effects Analysis (FMEA) and Hazard and Operability Analysis (HAZOP), is the perspective from
which potential failures are determined. With conventional techniques, the process of failure
prediction proceeds linearly from an articulation of the system's function(s) to what may occur if
there is a failure (absence) in delivering these functions. In other words, the analytical line of
reasoning follows design intent. Given a potential failure, the effect of the failure, the probability
that it will occur, and the ability to detect it are determined. Once these parameters are quantified
(often very subjectively), a calculation of risk is made. If the risk is determined to be
unacceptably high, changes in design or in detection capability can be suggested.
On the surface, the process sounds logical. There are, however, serious structural weaknesses
with these traditional approaches. The first weakness stems from the process used to determine
failures. The process of failure determination is essentially a brainstorming exercise initiated by
probing what failures "might" occur. This process suffers from the same syndrome that the
original product design process is subject to -- psychological inertia. Also, because the analysis
of potential failures is accomplished within the same mental context that created the design in
the first place, there is a serious question of objectivity to be raised with this approach.
Engineers do not like to admit that their designs are failure prone. A second shortcoming of
traditional approaches is that the analysis of failures is based on intended or designed function.
The issue of "prohibited" functions is not considered. For example, the function of a handgun is
to shoot a bullet, and thus related failure analyses proceed along the lines of the original design
intent. The original designers did not intend to design a weapon used by children to shoot their
classmates; this prohibited function is not a part of conventional failure prevention techniques.
Additionally, to be more complete, functions must be analyzed not only from the absence of
intent, but also from the perspective of the function being performed insufficiently or excessively.
The most serious drawback of traditional approaches, however, is the absence of an integrated
problem solving mechanism to accurately pinpoint design deficiencies as a series of "inventive"
problems. An inventive problem is one characterized by an inherent conflict. Traditional
techniques do not make provisions for solving difficult technological problems in an inventive
way. An inventive approach recognizes system conflicts and attacks them head-on. In traditional
approaches, if the design is deemed to be too risky, correction of the problem is accomplished
through a number of design and redesign iterations or, as a stopgap -- redesign of the detection
systems. When the system deficiency is not defined as an inventive problem, the results are
often costly over designs, or the addition of auxiliary compensating systems making the original
design more complex.
All of the structural deficiencies noted above have been designed out of AFD. First of all, the
approach to determining potential failures is the reverse of the one used in conventional
approaches. In AFD, the power of the technique comes from the process of deliberately
"inventing" failures. The engineer has to transform himself or herself into a subversive. The idea
is to invent, cause and create failures. In the case of past failures, the analytical process
challenges one to invent a past failure. In future failure prevention, the logic proceeds along the
lines of inventing, creating or devising the most catastrophic failures conceivable.
In both instances, the engineer inverts the problem. The advantage to this approach is
analogous to a defense attorney becoming a prosecutor. The system's potential flaws are viewed
from a perspective that allows for full exploitation of a system's weaknesses. It is obvious that,
when all system deficiencies are made explicit, the team or individual can take more effective
countermeasures.
AFD also has an integrated problem formulation engine to fully exploit the power of TRIZ. Failure
prevention is transformed from a defensive to an offensive "inventive" exercise creating a
seamless process for failure determination and prevention.
The process is so effective that users will sometimes become disenchanted with their system as
having so many drawbacks that it is a wonder it will work at all. This is normal as these are
potential failures. It is incumbent on the technical analyst to prevent these from ever occurring.
Facilitating detection
Figure 1, below, depicts a typical FMEA document with Steps 2 through 9 called out. Reference
this figure to follow the discussion below.
Figure 1. FMEA
2. Define In the AFD process, this step starts with the definition of the
Functions Primary Useful Function (PUF) for the "ideal" system.
Ideation Failure Prediction -- to predict all dangerous or harmful side effects that might
be associated with a system, and find means of preventing them.
reliability engineers
students
anyone interested in improving his/her technical innovative skills and thought processes
The AFD® Failure Analysis software guides the user through the following process:
1. Document and analyze the system and failure using the Failure Analysis
Questionnaire.
to create a5. For each selected hypothesis, use the Problem Formulator
graphic model depicting the revealed root cause(s) of the failure; generate a set
of problem statements for each model.
The AFD Failure Prediction software guides the user through the following process:
1. Document and analyze the system using the Failure Prediction Questionnaire.
2. Use the Problem Formulator® to create a graphic model of the system, identify
the focal points by evaluating the system against a set of checklists, describe
the system’s relationships to its environment, and formulate inverted problem
statements for each focal point.
3. Use the I-TRIZ operators associated with each inverted problem statement to
generate failure hypotheses for the system and its external relationships.
5. For each scenario, identify the components required for it to be realized and
verify (using a set of checklists) whether the necessary resources are present.
Software modules
Both AFD® Failure Analysis and AFD® Failure Prediction software include the following modules:
Failure Analysis Questionnaire – a tool for identifying and documenting the root
cause(s) of a system or process failure.
Screen shot from the Innovation Guide module of the Ideation Failure Analysis software:
System requirements
16 MB RAM memory