TOOLS &ANALYSIS

METHODS. Lec16.
 TOOLS & ANALYIS METHOD.

• One of the significant challenges for organizations implementing SMS

programs is to acquire or train personnel who possess the skills and
knowledge in these disciplines, and an understanding of the tools and
analysis methods used in them.

• Several important tools have been discussed in other parts of this book and,
in the interest of brevity, won’t be repeated here. We’ll simply list the tool
and refer to the chapter where that tool is discussed.

• For some of the tools, an example is provided for its use; for others we refer
the reader to other sources. Three tools in particular Failure Mode Effects
Analysis, Fault Tree Analysis, and gap analysis are presented first and
receive a more extensive treatment than the other tools due to their value in
safety management in all organizations.
• Finally, a more extensive discussion is provided for two tools and methods
that should be highly valued as predictive safety management: Probabilistic
Risk Assessment and data mining.

• Tool.
• Failure Mode Effects Analysis.

• One of the tools that has been helpful in achieving this goal is failure mode
effects analysis (FMEA). The purpose of the FMEA is to reduce the risk of
failure. It accomplishes this by providing a systematic means of identifying
and assessing potential failure modes, understanding the root causes of
those failures, and determining the actions necessary to eliminate the
potential failures.
 The FMEA is a “bottom up” approach. The practitioner begins this process with
a list of potential failures, typically developed either through a detailed analysis
of the diagram of a system or the results of brainstorming potential failures with
experts on the system.
Often the most difficult aspect of employing the failure and reliability tools
discussed in this book is obtaining the appropriate values for reliability and
failure. Some sources for this data include;
Manufacturer’s data;
Industry consensus standards;
Simulation and testing;
Delphi estimates;
MIL standards;
Historical evidence;
Other industry-specific sources.
• Fault Tree Analysis.

• Fault Tree Analysis (FTA) was developed at the Bell Laboratories in the
early 1960s to evaluate the safety of an ICBM launch control system, and
was later adopted and refined by The Boeing Company, among others.
Today FTA is used by many organizations in their system reliability and
safety programs.
• FTA is a graphical tool for analyzing complex systems to determine potential
failure modes and the probabilities that these failures might occur. FTA can
be used for accident investigation purposes.

• The FTA is an event-oriented method and, because of this, it has several

advantages over other methods, including FMEA. Some of the strengths of
this tool relative to other methods of system analysis include:
• FTA has the ability to discover failure combinations that might not be evident
using other methods. This tool has the ability to combine human error
contributions and hardware failures in the same analysis, provided detailed
task analysis information is available. This feature makes the FTA a
powerful tool for the reliability (or safety) practitioner. FTA is by nature
proactive. It is a structured process that can help identify potential causes of
system failure before the failures occur.

• FTA leads to the understanding of all possible combinations of low-order

events that will cause a system failure. This is infinitely more important than
knowing the impact on the system of the failure of a single component.
Since FTA assumes the failure of the top event, and examines the lower
order events to determine the combinations that can lead to failure, it is
regarded as a deductive approach.
FTA is graphical, which makes it relatively easy to understand by all interested
parties, including management. FTA works with both quantitative and qualitative
analysis. This is an important feature since quantitative data is frequently
unavailable or unreliable for some safety events.
• Table 10.1 Commonly used gates in fault tree analysis.
• Table 10.2 Commonly used events in fault
tree analysis.
• There are only two gates used in this simple FTA—an AND gate (top), and
three OR gates. If the probabilities of failure of each component are known,
a quantitative analysis can be performed.
• Gap analysis.
• Gap analysis is a term used to describe a broad range of techniques to
examine and describe the difference between current conditions and
desired conditions. The purpose of the gap analysis activity will determine
the technique to be used. Generally, the steps to performing a gap analysis
include the following:

• Define the scope of the gap analysis.

• Review all information about the issue under study.
• Determine the criteria, standards, regulations, performance measures, or
other factors that represent the desired conditions.
• Possess a thorough understanding of current conditions.
• Determine the appropriate instrument upon which to record the differences.
• Perform the analysis.
• Record and document the results.
• Analyze the results.
• Take the appropriate action on the information gained as a result of the
analysis.

• Sometimes the determination of the desired conditions is rather

straightforward.
• As with many other techniques including auditing, planning a research
study, or planning a project, the preliminary work for performing a gap
analysis is often the most time-consuming and challenging step in the
process.
• As with many other techniques including auditing, planning a research
study, or planning a project, the preliminary work for performing a gap
analysis is often the most time-consuming and challenging step in the
process.
• A primary reference for many of the tools described below is the Federal
Aviation Administration Air Traffic Organization Safety Management System
Manual, Draft, Version 2.0, dated March 2006.)
• Accident/Incident Analysis.
• The purpose of an Accident/Incident Analysis is to use existing risk data to
prevent future accidents from occurring. There are many ways that this
analysis can be accomplished, but the objective is always to better
understand the current trends and causal factors associated with accidents.
• The factors associated with accidents and incidents are recorded in a
database, and then examined for trends and patterns, such as distribution
among subgroups, locations, time of day, activities occurring conjunction
with the accidents, and others.
Cause and Effect Tool.

Change Analysis. The purpose of Change Analysis is to look systematically at

possible risks situations where change is occurring. This tool examines the
differences from normal operations configuration, or activities that have
occurred due to planned or incremental changes. It is used for proactive hazard
and risk assessment in changing situations and during accident investigations
to understand changes that have occurred in the system.

Hazard and Operability Tool.

The Hazard and Operability Tool (HAZOP) is a brainstorming technique for

identifying hazards and operability problems at the completion of process
design or for planned modifications.
HAZOP is particularly useful for new operations where other methods that rely
on experienced operational personnel are less effective. HAZOP is regarded as
the most highly structured of the hazard identification methods. The team uses
probing questions based on a series of standard guidewords to generate the
list of possible deviations.

The deviations are determined by combining the guideword with a variable

parameter or process term; that is:

Guideword + Parameter = Deviation.

Human Error Analysis. The purpose of a Human Error Analysis (HEA) is to
identify, analyze, and mitigate safety hazards associated with human error.
Everyone commits errors. HEA seeks to identify hazards related to human error
during the development of complex systems.
• . Interface Analysis.
• The purpose of an Interface Analysis is to uncover potentially hazardous
interfaces among seemingly unrelated activities. An Interface Analysis
examines potential energy exchanges between two activities to identify
potential hazards.

• Job Safety Analysis.

The purpose of a Job Safety Analysis (JSA) is to identify, analyze, and document
the steps involved in performing a specific job, to identify potential hazards
associated with the job, and to identify controls for the hazards identified
Job Task Analysis
A Job Task Analysis (JTA) identifies, analyzes, and documents human tasks within
a system, with the goal identifying hazards and risk factors requiring intervention.
A JTA analyzes each task in terms of physical and mental demands of the human
in the system.
• Management Oversight and Risk Tree. Management Oversight and Risk
Tree (MORT) is a comprehensive and detailed method for the analysis or
investigation of accidents and events. It uses a fault tree diagram (and FTA
symbology) where safety program elements are arranged in a logical,
orderly manner, beginning with the top event which is an accident or system
failure. It is then progressively broken down using the fault tree method
• Preliminary Hazard Analysis.
• The purpose of a Preliminary Hazard Analysis (PHA) is to provide a listing
of possible hazards that may be present in a system. As its name implies,
PHA provides a preliminary listing of hazards and, as such, is usually broad
but not particularly deep.
• Scenario Analysis.
The purpose of Scenario Analysis is to identify hazards by visualizing
scenarios, or possible outcomes, that could happen in an operation. Scenario
Analysis provides a structured means of viewing the flow of events in an
operation and considering the scenarios that lead to risk.
• What-If Analysis.
• A What-If analysis is a brainstorming approach that uses loosely structured
questions to identify hazards. The technique is typically used by a team of
experts with diverse backgrounds to generate a comprehensive review of
hazards.
• Predictive safety risk management: through modeling.
• Many of the tools and methods presented in this book have focused on
decomposing systems to better understand the component parts, In fact,
there are numerous methods that can be employed by SMS practitioners,
depending on the size and complexity of their organizations.
• Probabilistic Risk Assessment.
• There are two kinds of models used in problem solving and prediction in the
engineering world—deterministic and probabilistic. Many problems cannot
be solved using deterministic methods due to variability and uncertainty in
the dimensions of the problem; PRA is specifically designed to cope with
these uncertainties.
• NASA’s approach to a scenario-based
PRA process is as follows (NASA, 2002,
• pp. 10–13):
• Objectives Definition.
• Systems Familiarization.
• Identification of IEs—
• Scenario Modeling.
• Failure Modeling—
• The basic events and the intermediate failures are linked through logic
gates, such as AND and OR gates.
• Data Collection, Analysis, and Development—
• Quantification and Integration.
• Uncertainty Analysis—
• Sensitivity Analysis.
• Importance Ranking.
• Developing these scenarios is not a trivial undertaking, and locating the
input data for the model can often be the most difficult part of the process
• . Technique for Human Error Rate Prediction (THERP). THERP is a
technique for predicting human error probabilities and system degradation
based on human errors.
• THERP is a technique for predicting human error probabilities
• and system degradation based on human errors.
• THERP is a technique for predicting
human error probabilities and system
degradation based on human errors.

• Define the system or process.

• Identify and enumerate all human operations performed and their
relationships to the system.
• Predict error rates for each human operation.
• Determine the effect of human errors on the system.
• Make changes that will reduce the system failure rate.
• PRA has been used in numerous safety applications where safety is a
function of interactions between multiple factors
• Monte Carlo.

• Perhaps the most widely used probabilistic method, Monte Carlo methods
are popular in many disciplines of science and in many industries. Monte
Carlo Analysis (MCA) uses computer simulation to combine multiple
probability distributions in a risk equation.

• Depending on the objective of the risk analysis, risk modeling typically

provides the following outputs;

• Estimates of risk at predetermined levels of confidence A risk profile that

ranks risk events in order of decreasing risk values. An exposure profile
indicates the range of consequential costs for risk events.
• Data Mining.
• Data mining (DM) has become an increasingly popular analytical method in
numerous industries, including science, finance, homeland security,
business and marketing,transportation, and nearly every other industry.
• Those applying DM methods seek to make greater use of existing
databases to learn more about the problem or issue at hand than more
traditional methods have afforded. Given the vast data collection efforts that
are occurring in aviation safety, DM has the potential to unlock secrets
hidden in that data.
• What is Data Mining?

• Data mining is an analytic process designed to explore large amounts of

data in search of consistent patterns and/or systematic relationships bet
• ween variables (StatSoft, 2003).
• Regression modeling normally begins with a hypothesis which is tested by
this common statistical technique.
• is an important concept in DM. Through the study of multidimensional
Visualization graphs the analyst is able to detect trends, patterns, or
relationships.
• Cluster analysis is an exploratory data analysis tool that consists of several
different algorithms and methods for grouping objects of similar kind into
respective categories.
• Steps in DM
• Crucial concepts in DM.
• Data preparation, cleaning, and transformation.
• Feature selection.
• Feature extraction.
• Predictive DM.
• Sampling, training, and testing (hold-out) samples.
• Over-sampling particular strata to over-represent rare events (stratified
sampling).
• Machine learning.
• Deployment.

• Finally, STATISTICA’s Intelligent Problem Solver (IPS) is a sophisticated tool

for the creation and testing of neural networks for data analysis and
prediction problems.
• THANK YOU.