Professional Documents
Culture Documents
Mustafa Alshekh
Roland Jochem
Summary
Global competition has increased the complexity of products and their supply chains, which forces
companies to use more reliable risk analysis methods to minimize the risk of potential failures in their
products and their manufacturing processes. This paper presents a further development of the cross
impact based FMEA (CFMEA) method. CFMEA is a method to reduce the risk of potential failures in
the manufacturing of new products and to eliminate the weaknesses of conventional FMEA. It is a
combination of classical FMEA, scenario technique (cross impact analysis) and cost of failures and
provides an accurate assessment of the potential failures in a system, which consists of a product and the
necessary processes to manufacture it. In the CFMEA the influence of the detection distance on the cross
impact and on failure costs will be also analyzed. It is capable to determine the impacts of a failure in the
manufacturing process not only on the main process but also on the related functions in the product
(both elements and system level). It uses also advanced cross impact analysis to quantify the cross
impacts between the failure causes in the process and to measure the effect of detection distance on the
cross impact. It provides the opportunity to calculate the cost of failures in the manufacturing process in
order to evaluate the profitability of planned optimization actions.
Keywords
Failure Mode and Effects Analysis (FMEA), Advanced Cross Impact Analysis, Cost of Failures
1. Introduction
Today, the global competition requires the development of customer-oriented products in a short time
(Dippe 2008). Therefore companies must rely more on their suppliers to fulfill the requirements and
expectations of their customers. But the implementation of this out-sourcing strategy is associated with
risks. A failure within the supply chain has impact on the product quality, cost, and time to market which
means risks to the company. As a strategy for risk identification and risk minimization the methods of
quality management can be used, e.g. failure modes and effects analysis (FMEA), fault tree analysis
(FTA), cause and effect diagram, poka-yoke as well as reliability techniques (Geiger and Kotte 2008;
Tietjen and Mller 2003). Since the fact that FMEA is the most used method for risk assessment in the
companies (Dittmann 2007), we will concentrate here on developing this method.
As an extension to the first publication in this issue (Alshekh and Jochem 2012) we intend in this paper
to include the impact of the detection distance on the cross impact between the failure causes.
Section (2) provides an overview of the classical FMEA. Section (3) explains the approach of CFMEA
while the important steps of the CFMEA are explained in section (4) using case study. Section (5)
discusses the results.
Page 1 of 15
2. FMEA-OVERVIEW
2.1. FMEA Basic Concept
FMEA is a formalized analytical method, which defines and avoids the risks associated with
manufacturing of products and processes systematically and completely (Dietze and Mockenhaupt
2002). It was developed at NASA during the development of the Apollo project in the 60s (Kamiske
2009). The first utilization of the FMEA in Germany was in 1980 and was called Ausfalleffektanalyse
according to DIN 25448. The classical FMEA method will be performed according to the VDA 4.3 as
the following:
1. Definition of the objectives of the FMEA, definition of the system under consideration,
coordination of interfaces with other systems or subsystems and building a system structure.
2. Assigning functions to each element in this structure and form functional nets.
3. Defining the malfunctions of each function and form failure nets.
4. Determining the severity of the failures sequences and the occurrence probability and detection
probability of failure causes under consideration of the current state of prevention and detection
actions and calculating the risk priority number (RPN).
5. If required a particular extra actions for certain failure causes will be defined to minimize the risk
of this failure causes to an acceptable level. The risk minimization will be carried out through
minimizing the occurrence probability of failure causes, increasing the probability of detection of
failure causes or minimizing the severity of failures sequences (Pfeufer 2002; Werdich 2011).
2.2. FMEA State of the Art
Many authors have attempted to adapt the classical FMEA or to combine it with other methods to
improve its efficiency and effectiveness. Advanced FMEA (AFMEA) was developed to identify and
avoid the failures early in the design process by using system behavior modelling (F. Eubanks, Kmenta
and Ishii 1996, F. Eubanks, Kmenta and Ishii 1997; Kmenta, Fitch and Ishii 1999; Kmenta and Ishii
1998). Expanded FMEA (EFMEA) has provided a solution for two important problems in classical
FMEA: prioritizing the risk priority number (RPN) and the comparison between the optimization actions
(Bluvband, Grabov and Nakar 2004). Haffner 2005 invented a model to calculate the potential benefits
by implementing a Process-FMEA (Haffner 2005). Neghab 2011 has combined qualitative methods of
risk analysis in terms of FMEA with the quantitative methods in terms of discrete event simulation
(DES) to create a new framework for risk analysis in the manufacturing processes (Pirayesh Neghab et
al. 2011). A Failure - Process - Matrix (FPM) has been developed at the fraunhofer institute for
manufacturing engineering and automation (IPA) in collaboration with a leading original equipment
manufacturer (OEM) in the automotive industry for a comprehensive analysis of complex assembly
processes e.g. motor assembly (Werdich 2011).
3. CFMEA Methodology
3.1. Basic Concept
We have seen in the last section that the consideration of the entire system (product-process) with the
cross impacts between its elements is currently not the case. The objective of this paper is to develop a
risk analysis method based on a FMEA considering cross impact. The CFMEA will offer the
opportunity:
Page 2 of 15
To identify and quantify the cross impacts between main failure causes in the process - FMEA using
advanced cross impact analysis and integrating them into the risk assessment process. So the
CFMEA offers the developer the opportunity to compare between several manufacturing and
assembly technologies regarding of the possible cross impacts between their failure causes in the
process and to choose the one with minimum risk, which mean that CFME can be used as a
development tool early in the conception phase of product development.
To identify the impact of the failure causes not only on the process but also on the related functions
in the product. This means that the severity of the failure sequence will be measured not only on the
top process level but also on the product level (Element level & System level)
To predict the cost of internal and external failures in the process.
To make a decision regarding optimization actions with taking into account the cross impacts
between the failure causes and their costs.
environmental conditions, test reports, warranty data, engineering drawings, failure rate data, and
organizational information, e.g. sequence of meetings, roles of team participant, required additional time
for CFMEA. In addition all experts who worked on the creating the classical FMEA must work
afterwards on CFMEA. Definition phase consists of several steps which must be performed parallel and
sequentially. We assume that all information about the planned production volume (N), acceptable level
of Risk Factor (RF), the manufacturing costs, planned added value and planned profit in each process
step, the scrap rate for each failure cause and the current status for prevention and detection actions are
available. Following characteristics must be defined during definition phase:
1. The rating tables for the occurrence probability (O), detection probability (D), and the effect level
(S2 & S3).
S2: weighting of the impact of failure cause in the process on the related functions in product (element
level).
S3: weighting of the impact of failure cause in the process on the related system functions in product
(system level).
Rating Number
Evaluation
No rating
Imperceptible
3-4
Low
5-6
Moderate
7-8
High
9 - 10
Very high
on this definition. In case some critical failure causes would not be considered, this would have a
negative effect on the accuracy of the results. At the same time considering many failures causes
which are not critical, would lead to an increase in the complexity and the time needed for the
analysis and optimization phases. The main failure causes can be determined through compromise
between experts or using the four phase model of Haffner (Haffner 2005; Jochem 2010).
Compromise between experts can be achieved through discussion and the use of quality tools Q7
(e.g. brainstorming) and management tools M7 (e.g. relations diagram) (Kamiske 2009).
4. The impact of each failure cause on the related function in product (element and system level). Data
input for this step are Product- and Process- FMEA
5. The weighting (S2 & S3) of previous impacts
3.3. Analysis Phase
This phase is designed to create a concept for quantification of the cross impacts between the main
failure causes in the process based on the information provided in the definition phase and to use this
concept in order to calculate the risk factors and the cost of each failure cause.
In principle the analysis phase consists of three main steps:
1. Advanced cross impact analysis.
2. Calculation of the new probability of occurrence and risk factor for failure causes with consideration
of the cross impact.
3. Calculation of the failures costs.
3.3.1. Cross Impact Analysis
We have used in the last paper the so-called "paper computer"(Vester 2003) to determine the cross
impacts between the failure causes systematically. This method belongs to the systematic formalized
scenario techniques (Kosow and Gassner 2008), and it has been used in many researches to make a
quantifiable prediction for the cross impact between different variables in a system; see (Giebel 2010).
Four types of failure causes can be discriminated according to the matrix of cross impact analysis:
Active failure causes, which have a major impact on the other failure causes and are little impacted
by them
Passive failure causes, which have little impact on the other failure causes, but are strongly impacted
by them
Dynamic failure causes, which have a major impact on the other failure causes and are
simultaneously impacted by them
Buffer failure causes, which have little impact on just a few other failure causes and are
simultaneously impacted by the others
In this paper we intend to involve the influence of detection distance on the cross impact, therefore we
have developed the cross impact analysis matrix of Vester. The advanced cross impact matrix is able to
quantify the cross impact between the failure causes while taking into account the detection distance of
this failure causes, see the next paragraph.
Page 5 of 15
In time detection: Is the probability that the failure cause when it occurs, it will be detected in the
same process step through particular mechanisms. Normally, the detection probability in the
classical FMEA refers to the detection actions in the process without concentration on the
detection distance or place of detection.
Later detection: Is the probability that the failure cause when it occurs, it can be also detected in
another process step (detection of slip), e.g. poka-yoke, sensors etc.
End of line detection: The failure cause will be detected before it appears by the customer. Here
help this type of detection to calculate the internal and external failure costs accurately, e.g.
functional test or pneumatic test in the production of valves.
No detection: the failure will appear or it will be detected by the customer. In this case the failure
will be classified as external failure and the resulting costs can be quite high.
Impact on
Impact of
FC11
FC12
FC13
FC21
FC22
Negative Sum
FC11
0
0
0
1
1
FC12
FC13
FC21
FC22
Active Sum
1
2
D
1
2
0
1
1
D
4
4
4
5
4
21
1
2
0
6
3
2
3
1
8
Page 6 of 15
2. Passive and dynamic failure causes are considered as measurement failure causes, since they are
strongly impacted by the others. In addition, they will be used later in the optimization phase to measure
the effectiveness of the optimization actions.
3. Buffer failure causes are considered as stable failure causes because they play a small role in the cross
impact.
This new characterization of the failure causes has two advantages:
Only optimization failure causes that are not detected will be considered in the cross impact.
Since the impact of the passive variables is very low, it will not be considered in the calculation
of the new probability of occurrence. In addition it will be assumed that the rounding error in the
calculation is equivalent to the impact of the passive variables.
It is necessary also to point out that we are dealing here with the occurrence probability, which
exists in classical FMEA, therefore the calculation roles of the normal and conditional
occurrence probabilities cannot be applied.
: Increase in the probability of occurrence of the failure cause due to cross impact
: Increase of the probability of occurrence level due to the cross impact. It is derived from
with using the rating table for ,
[1, 10]
: New occurrence probability level of the failure cause,
Page 7 of 15
: Occurrence probability level of the failure cause without consideration of the cross impact,
: Number of process steps. : Number of failure causes.
cause,
FMEA.
: Occurrence probability of the failure causes that have an impact on the investigated failure
the values of
can be received from the rating table for und in the classical
4. Calculation of the Risk Factor (RF), where RF is obtained from the product of these three
indices:
secures that the impact of the failure cause will be measured on the
: Cost of
Page 8 of 15
CCRI(total): Total internal constant rework costs, CCRI: Individual constant rework costs.
The Conditional Costs: Are the costs of possible reduction in value. These occur only if the reworked
unit has less value than planned because of failure.
: Number of rework (intern), : Planned value of the product at this time with consideration of the
profit (intern),
: Depreciation due to the rework with consideration of the estimated profit
The Variable Costs: Are the costs that result from the consequences of rework. They include:
These costs are difficult to detect and vary from company to another; therefore, these costs can be only
estimated by comparison with other companies, which had a comparable problem or by comparison with
the company records from previous years.
Cost of Scrap: These include constant and variable costs.
The Constant Costs: Include the cost of personnel, material, downtime, scrapping, and other potential
costs, which depend on the company and product.
CCSI(total): Total internal constant scrap costs, CCSI: Individual constant scrap costs.
The Variable Costs: Are the costs of the quality related quantity deviation, these costs consist of
measurable cost (cost of the unrealized profit) and non-measurable costs (cost of the loss of customer
confidence), which is the reason of lost future contracts. Normally, these costs are difficult to measure
because they depend on many not quantifiable factors (Market, Regularly roles, Customer itself). So
these costs can only be estimated.
: Cost of unrealized profit,
Page 9 of 15
The conditional costs: Are the costs of possible reduction in the product value and they will be
calculated in the same way as is done for internal rework costs.
The variable costs: are the same variable costs in the internal rework costs.
The cost of scrap: These consists of constant costs and variable costs
Constant costs: Consists of the cost of manufacturing, transport and recall actions, scrapping, marketing,
product liability, and other potential costs, which depend on the company and the product.
Variable costs: Consists of the cost of the possible loss of the customers and the cost of image loss.
These costs can be estimated through comparing with other companies.
3.4.Optimization Phase
The decision about the optimization actions (prevention and detection actions) is made in this phase.
This decision should have an impact on the entire system (all failure causes and their cross impacts) and
lead to minimization of the risk factor for all significant failure causes to the acceptable level. This phase
consists basically of five successive steps.
Calculation of Cross Impact Intensity Number: This step is designed to calculate the cross impact
intensity of the optimization failure causes in order to minimize the calculation time needed in the
measuring of the system sensitivity.
{
Prioritization of the Optimization Failure Causes: In this step the optimization failure causes will be
prioritized regarding their cross impact intensity number.
Definition of the Possible Optimization Actions: A list will be performed for all possible optimization
actions (prevention and detection actions) for the optimization failure causes. The definition of each
optimization action should include the following elements: Description of the optimization action, its
effect on occurrence or detection, potential reduction in O or D after implementation and
implementation costs.
Analysis of System Sensitivity: The system sensitivity will be measured for each optimization action
while taking into account the prioritization of the optimization failure causes. Here in compliance with
the prioritization of the optimization failure causes an optimization action will be implemented and its
impact on the measurement failure causes will be measured.
Decision making: The optimization actions, that are profitable (Benefit > Cost) and safe by minimizing
the risk of failure causes to the acceptable level, must be implemented.
The steps of the optimization phase will be end, when all risk factors are below of the acceptable level.
4. Case Study
We will try to explain the main steps of the CFMEA by means of an example about metallic
constructions. See figure 2. The column set shown in the figure 2 is the main element in the field of
metallic constructions. The beam profile is usually H or I beam. We are going to explain the CFMEA
analysis of the manufacturing of the column set. The manufacturing process of the column set consists
mainly of:
Page 10 of 15
Manufacturing the beam (in this case H profile), manufacturing the upper and lower plate, assembly the
beam with the upper and lower plate through welding.
Page 11 of 15
Description of Impact
Correction Factor E
Great impact
80%
Intermediaries impact
60%
Low impact
40%
No impact
0%
P(O)
P(D)
P(ND) PCI(O)
O1=6
0.005
95%
5%
O2=5
0.002
97.5% 2.5%
2.5*10-4
0.5*10-4
Page 12 of 15
Calculating the risk priority number (classical FMEA) and risk factor (CFMEA) for the failure cause
FC3
So due to cross impact (increase in the occurrence probability of the failure FC3 from 3 to 5) and
evaluating the impact of the failure cause FC3 not only on the process level but also on the product level
(S`=7) there is a need for an optimization action. But according to the results of prioritization of the
optimization failure causes Wrong beam (wrong parameter) is taken must be first optimized.
Applying a specific sensors system is the optimization action of the failure cause FC1 (Wrong beam
(wrong parameter) is taken). This optimization action costs (20.000) and reduces the probability of
occurrence of the failure cause FC1 to (2).
Now when O1=2, O3 will be automatic reduced to O3=4. And the Risk Factor of FC3 in this case
RF=112<125.
So the reduction of the probability of occurrence of FC1 has reduced the risk factor of the failure FC3 by
(112) below the acceptable level of (125). Since (profit>cost) the implementation of this optimization
action is profitable and we need not any more optimization action for the failure cause FC3.
5. Conclusion
With this new approach of risk analysis companies will be able to assess the risk of the possible failures
in their products and processes while taking into account the possible cross impacts between them.
Moreover this approach provides a method to support the decision making process during the planning
for the optimization actions by calculating the cost of failures and the application of costs-benefits
analysis. So the CFMEA can in compare with classical FMEA:
Page 13 of 15
All these aspects make CFMEA more able to be an effective and data-based analysis method.
6. Bibliography
Alshekh, Mustafa, and Roland Jochem. The three stages model for optimizing the failure mode and
effects analysis (FMEA): Cross impact based FMEA (CFMEA). Quality - Access to Success 2012
(2012): 53.
Bluvband, Zigmund, Pavel Grabov, and Oren Nakar. Expanded FMEA (EFMEA). Reliability and
Maintainability Annual Symposium RAMS (2004): 3136.
Dietze, Adrian, and Andreas Mockenhaupt. Werkzeuge des Innovationsmanagements. Berlin:
dissertation.de - Publisher in Internet, 2002.
Dippe, Andreas. Einsatz von Qualittstechniken in der Entwicklung komplexer Systeme: Entwicklung
eines Vorgehensmodells am Beispiel der Automobilindustrie. Aachen: Shaker, 2008.
Dittmann, Lars. OntoFMEA: Ontologiebasierte Fehlermglichkeits- und Einflussanalyse. Wiesbaden:
Dt. Univ.-Publisher, 2007.
F. Eubanks, Charles, Steven Kmenta, and Kosuke Ishii. System Behavior Modeling as a Basis for
Advanced Failure Modes and Effects Analysis. The 1996 ASME .
F. Eubanks, Charles, Steven Kmenta, and Kosuke Ishii. Advanced Failure Modes and Effects Analysis
Using Behavior Modeling. The 1997 ASME.
Geiger, Walter, and Willi Kotte. Handbuch Qualitt: Grundlagen und Elemente des
Qualitatsmanagements: Systeme - Perspektiven. 5th ed. Wiesbaden: Friedr. Vieweg, 2008.
Giebel, Mechael. Wertsteigerung durch Qualittsmanagement., Dissertation (2010).
Haffner, Andreas. Ein Modell zur Bestimmung der monetren Einsparungspotenziale bei der
Durchfhrung einer Fehlermglichkeits- und Einflussanalyse (FMEA). (2005).
Jochem, Roland. Was kostet Qualitt? Wirtschaftlichkeit von Qualitt ermitteln. Mnchen: Hanser,
2010.
Kamiske, Gerd F. Handbuch QM-Methoden. Mnchen: Hanser, Carl, 2009.
Kmenta, Steven, Peder Fitch, and Kosuke Ishii. Advanced Failure Modes and Effects Analysis of
Complex Processes. ASME (1999).
Kmenta, Steven, and Kosuke Ishii. Advanced FMEA Using META Behavior Modeling for Concurrent
Design of Products and Controls. ASME (1998).
Kosow, Hannah, and Robert Gassner. Methoden der Zukunfts- und Szenarioanalyse. Berlin: IZT, 2008.
Lin, Gerhard. Qualittsmanagement fr Ingenieure. 3th ed. Mnchen: Hanser, Carl, 2009.
Masing, Walter, and Jrgen P. Blsing. Handbuch Qualittsmanagement. 4th ed. Mnchen : Hanser,
1999.
Pfeufer, Hans-Joachim. Fehler-Mglichkeits- und Einfluss-Analyse (FMEA). Dsseldorf: Symposion
Publishing GmbH, 2002.
Pirayesh Neghab, A., A. Siadat, R. Tavakkoli-Moghaddam, and F. Jolai. An Integrated Approach for
Risk-Assessment Analysis in a Manufacturing Process Using FMEA and DES. Quality and
Reliability (ICQR), 2011 IEEE International Conference (2011)
Page 14 of 15
Tietjen, Thorsten, and Dieter H. Mller. FMEA-Praxis. 2th ed. Mnchen: Hanser, 2003.
Tomys, Anne-Katrin. Kostenorientiertes Qualittsmanagement: Ein Beitrag zur Klrung der QualittKosten-Problematik. Mnchen: Hanser, 1994.
Verband der Automobilindustrie VDA. Produkt- und Prozess-FMEA. 2006, Nr. 4.
Vester, Frederic. Die Kunst vernetzt zu denken: Ideen und Werkzeuge fr einen neuen Umgang mit
Komplexitt; ein Bericht an den Club of Rome. 3th ed. Mnchen: Deutscher Taschenbuchverlag,
2003.
Werdich, Martin. FMEA - Einfhrung und Moderation. 1st ed. Wiesbaden: Vieweg + Teubner, 2011.
Page 15 of 15