Sie sind auf Seite 1von 6

Risk Assessment Using Design Review Based On Failure Mode

Roland Schmidt, PhD, ABB Switzerland


Gernot J. Riedel, PhD, ABB Switzerland
Klaus Kangas, ABB Oy Drives
Key Words: DRBFM, FMEA, Risk Assessment
SUMMARY & CONCLUSIONS
In order to improve sustainability of industrial
applications, big efforts are underway to increase efficiency of
electrical power transmission, power conversion, or power
usage. The availability of electrical converters used in these
applications is crucial for high efficiency. Therefore, risk
assessment has to be performed during product development
to improve their reliability. The method typically used is
Failure Mode and Effects Analysis (FMEA). In this paper
another method called Design Review Based on Failure Mode
(DRBFM) is applied to a typical power converter and
evaluated. The DRBFM approach used was found to be much
more efficient, if the focus lies on the identification of the
critical problems and on improvements. On the other hand,
FMEA is the method of choice for focusing on a complete list
of possible failures.
1 INTRODUCTION
Improving sustainability is a key goal in the development
and enhancement of electrical power generation, power
transmission, and electrical machines [1]. Big efforts are
under way to reduce losses, to increase efficiency, and to
improve the stability and availability of electrical power
transmission. The interconnection of power systems with
high-voltage direct current (HVDC) technology and the use of
flexible AC technologies (FACTS) reduce losses. Electrical
converters optimized for high efficiency increase the
sustainability of photovoltaic power plants or wind parks.
Variable speed drives for motors also reduce the waste of
electrical energy [2] [3].
For the overall efficiency of sustainable energies in
general and wind power plants in particular high availability is
crucial. Therefore the reliability of the wind turbines and their
components is highly important [4]. This applies especially
for offshore wind park installations, since maintenance and
repair of equipment need high logistic effort, depend on
weather conditions, and are consequently very expensive [5].
Electrical equipment is the main contributor to system failures
of wind parks [6]. In order to increase the service and
maintenance intervals and to reduce instant failures, improved
reliability of the electrical equipment is necessary.
To reach this goal, quality and reliability methods have to
be introduced into design and production of large electrical

978-1-4244-8855-1/11/$26.00 2011 IEEE

equipment. Automotive industry faced a similar situation,


when it started to ask for zero defect products and zero defect
strategies [7]. This goal cannot be achieved by quality control
and reliability testing alone. Knowledge and understanding of
possible failures and their mechanisms are necessary. This
resulted in physics-of-failure approaches [8] and design-in
reliability [9]. Even more important became risk assessment
methods which identify critical or key failures in a structured
way.
Such risk evaluations should be included within the full
production process of products. Already at the early design
steps reliability issues and possible failures should be
addressed. The later a failure is detected the higher are efforts
and costs to eliminate it. A rule of thumb states that with each
step during the product development process the costs of
failures increase by a factor of 10 [10].
Several approaches for risk assessment during
development and production are available, including Failure
Mode and Effect Analysis (FMEA), Fault Tree Analysis
(FTA), Ishikawa diagram, etc. They follow a strict procedure
and are highly structured. Whereas FTA and Ishikawa
diagram start from an observed or expected failure, FMEA is a
more general approach trying to identify all possible risks.
FMEA [11] is the most widely used tool to assess risks of
products. It is a team-oriented and cross-functional method
used for failure prevention. The result of an FMEA is a
thoroughly filled-in form (Fig. 1).
However, there are some serious drawbacks in performing
an FMEA. It is very time consuming and binds a lot of
engineering and management manpower. Participants are
often not really interested to gain the best benefit, but want to
get over and done with it as soon as possible. Especially in
the early development stages, FMEA is often not performed
due to a lack of design information, which is needed to fill in
the FMEA form thoroughly. Additionally, if an FMEA is not
reviewed and updated regularly, it will loose its main
capability, namely giving an actual overview of risks and
taken countermeasures for improvement.
Instead of
identifying possible failures and defining countermeasures,
time and energy is wasted to discuss intensively the ratings for
severity, occurrency, and detection of failures. Due to these
facts, some reluctance is often observed against the
implementation and performance of FMEA within a company.

Fig. 1 - Example of an FMEA form.


General details on this method, and especially on the rating
problem, can be found in [12] [13]. Most FMEAs focus on a
complete filled-in failure list. This is meanwhile often done
with software tools. Improvement will not be achieved by
perfectly filled-in forms, but through actions derived from an
FMEA.
To overcome some of these drawbacks, the FMEA
method is changed or adapted to the internal needs and
experiences of a company or product, e.g. [14]. Other
methods derived from FMEA may come up and will be
established within industrial sectors. For example, Design
Review Based on Failure Mode (DRBFM) is one of these
methods, which is often used in the automotive industry.
2 DRBFM METHOD
Design Review Based on Failure Mode (DRBFM) was
developed and established within Toyota Motor Corporation
by Tatsuhiko Yoshimura [15]. He realized that often the
FMEA table is blocking a creative design review. As changes
involve the highest failure potential, DRBFM was developed
mainly to identify failures due to a design change.
Background is the concept of Mizen Boushi (reliability

problem prevention) and the innovation process GD3 (G.D.


Cube) [16]. GD3 stands for
good design
good discussion
good design review
Good design means that parts should be kept, if they
already have a proven good design and if they do not have to
be changed necessarily. Good discussion stands for creative
and open minded discussions during which risks and failures
can be revealed without any reservation. It is the difficult task
of the leader or moderator to establish such a good discussion.
Good design review means, that all design modifications are
critical and are in depth analyzed and discussed.
'DRBFM is a method of discovering problems and
developing countermeasures by taking notice of and
discussing intentional changes (design modifications) and
incidental changes (changes in part environment). The
method encourages the human ability to find problems, and is
a practical tool based on FMEA and FTA' [16].
DRBFM consists of two parts. In the first part, the
analytical phase, the designers put together the changes and
identify possible failures for the function and implementation
related to them. They also define countermeasures. In the
second part, the design review, an interdisciplinary team
reviews the design and the failures identified by the designers.
Additional failures will be identified and actions defined [16]
[17]. To summarize and structure the results a form derived
from FMEA form is used (Fig. 2). The filled in form should
give a thorough overview of components (column 1), the
affected functions and requirements (column 2), possible
failures or loss of function (column 3), the root causes of the
failure (column 4) and the impact or effect of a failure
(column 5). A small column "Evaluation" belongs to column

Fig. 2 - Headline of a DRBFM form. The light fields should be prepared by the design team.
The dark fields will be filled in during the design review meeting [17].

5. In this column it is possible to rate the effect (severity) or


to identify it as hazardous or critical. Measures already taken
in the design phase or production to eliminate the failure root
causes or to detect them are given in column 6. Three
columns are reserved for corrective actions. These actions
will be assigned to the design phase (column 7), the test or
verification phase (column 8) and the production phase
(column 9). Each action is assigned to a responsible person
together with a deadline. Results and follow up results can
later on be put in column 10. Especially the question 'Any
other concern?' and 'Any other causes?' should lead to a deeper
thinking on failures and root causes. For very important
failures FTA, 5-Why method or Ishikawa diagram can be used
to identify the real root causes.
Fig. 3 shows a DRBFM from during the discussion
process [17]. Removable sticky notes are used instead of the
fixed fields of a table. This allows a more creative and lively
discussion. The focus is not any more on the task of filling in
the blank fields of the table, but on the subject under
discussion.

at to perform risk assessments with the focus on identifying


possible failures and risks and to achieve product
improvement. As described, FMEA shows some drawbacks,
if used as tool. Here DRBFM was chosen to be evaluated
during risk appraisal. To perform a DRBFM of power
converters, a team was assembled consisting of several very
experienced product engineers. The aim was to identify risks,
evaluate them, and define possible actions for improvement.
The assessment was not performed as a functional or system
risk estimation. Instead, it was looked at components, groups
of components, design, and manufacturing. Therefore, the
converter was divided in several main parts (Fig. 4). Board
assembly was dealt with as a separate part, as it is related to all
printed circuit boards.

Fig. 4 - Principal parts of a converter.

Fig. 3 - Typical picture of a DRBFM table


The sticky notes can easily be grouped and regrouped
during the team session supporting an open discussion.
Failure (3), failure root cause (4), failure effect (5),
countermeasures (6) and recommended actions (7-9) do not
have to stay necessarily within one line. If necessary or
wished relations can be indicated alternatively with numbers.
The most important result is the list of recommended actions,
which gives the input to improve the quality and reliability of
the product.
3 APPLICATION
The reliability of power electrical converters is very
important as discussed above. Therefore methods were looked

The main focus of this risk assessment was not to put


together all possible risks, but to identify main failure
possibilities, to investigate their root causes, and to define
possible actions to eliminate these failures. This risk
evaluation was not part of a product development with design
reviews. Therefore, only the first part of the DRBFM was
used as a structured way to sum up and to order the upcoming
ideas. The form was modified, as it is shown in Fig. 5.
At the beginning of the assessment, a short introduction to
the method was given. For the DRBFM no software tool was
used, but boards with plain paper. The entries were written on
sticky notes, pinned to the board, and freely placed during the
discussions. For each part of the converter an own DRBFM
table was used. These main parts were then divided in several
electrical, mechanical, or optical components or component
groups.
First, the main function of the group under investigation
was defined. Afterwards, possible failures were collected and
the root causes identified. It was not looked at the loss of
function in detail, Instead it was assumed, that the loss of
function is a failing converter no matter whether the converter
just stops or the power module explodes. The analyses of the
different main components were photographed for
documentation.
Fig. 5 shows two examples of the numerous entries
achieved during the discussion. One example is the thermal
paste (1). It is applied between the power module and the
heatsink. Its function (2) is to transfer the heat from the power

module backside to the heatsink to cool the module. If the


heat transfer is not good enough the power module will heat
up during operation (3). This will lead to a power module
failure (5). There are several root causes (4) given: too much
thermal grease, too little thermal grease, or bad material (not
showing a low thermal resistance). There are already some
tools implemented (6) to guarantee the right amount and
thickness of the thermal grease. As a possible action for
design (7), it was defined to look into materials with improved
thermal conductivity and applicability. No entries were made
during this DRBFM for evaluation (5). Instead important
entries were marked with an up-arrow and less important
entries with a down-arrow (see column 4). Also no entries
were made in columns (8), (10), and (12). The column result
or follow-up (13) was kept empty as no decisions were made
during the assessment.
The second example is taken from board assembly (1).
The loss of function (3) looked at is simply that it is not
working. One root cause (4) out of many is that a wrong
component is used. A sequence of root causes was identified.
Components were mixed during board assembly. This may
happen, if the reels containing the devices are mixed or taken
falsely as they sometimes are not barcoded.
Wrong
components may be detected during the performance of
electrical tests (6), however not all wrong components might
be detected and there might be additional cost for rework.
Therefore as a possible action concerning production (11) a
request to the board assemblers (often external suppliers) is
attached to use barcoded reels for right identification. This
example shows that it is very easy to group and display a
sequence of root causes.
The results in the tables were evaluated and summarized
in a report for better understanding. They can be discussed
with other engineers and management defining next actions.

Such a discussion may take over the role of the design review.
4 EVALUATION OF THE APPROACH
After the risk assessment the method and the approach
were evaluated and compared with the standard FMEA
method. The following advantages and disadvantages of the
DRBFM approach were found.
The advantages are:
Very time efficient.
Possible problems are identified and listed.
It gives a good overall picture. It permits an overview of
possible failures and gives a list of possible actions:
A clear overview of topics to be fixed is the result. The
result of the assessment is not a filled-in table, but
possibilities for improvement.
All possible actions address realistic risks.
The focus was on problems not on rating or calculating
the risk priority number (RPN). Therefore:
No time was wasted for rating.
The results and actions can directly be used for new
product development.
It looks like brainstorming, but it is not. Brainstorming
has several drawbacks [14]. Therefore it was very
positively registered that it is a different method. The
approach is much more structured, makes it easier for the
moderator to guide the discussion and leads to wellthought root causes and therefore possible actions.
The disadvantages are:
The information of a risk priority number (RPN) is
lacking. A ranking of failures and actions is not so easily
possible.
Failures and especially some root causes are not taken
into account, if they were assumed as not so important by
team members.

Fig. 5 - Adapted DRBFM form with some example entries of the risk assessment.

The following ideas for future use were collected:


The function of the component discussed should be more
in the center of the discussion. From there the failures
and the root causes can be derived easily.
The 'ideal team' will consist of:
2-3 product (development) engineers,
2-3 predevelopment engineers, and
a moderator
The following skills and information were used or
needed:
product example on display
schematics
design know-how
knowledge of failing parts
deep understanding of converter production and
testing / testing performed externally
information and background of involved suppliers
For this risk assessment the consumed time was about 10h
work in three sessions within 1 week.
The comparison between this DRBFM approach and a
typical FMEA was discussed and rated within the team. The
result is shown in table 1.

Overall Efficiency
Time Efficiency
Benefit (new ideas)
Feeling
Coverage
Standard /
Customer Request

DRBFM
4
5
3-4
4-5
3
2

FMEA
2
2
2-3
2
4
5

Table 1 - Comparison between DRBFM and FMEA


(from 1 = bad to 5 = very good).
The results in the table show clearly the benefits of the
DRBFM approach. Especially the time efficiency was rated
very high compared to a classical FMEA. This is mainly
because of the creative approach, the concentration on
problems and the omittance of detailed rating. Also the
benefit of new ideas was rated much higher. Regarding the
coverage of the possible failures the FMEA was rated better.
This is due to the fact that all failures are listed in the FMEA
table, whereas in this approach less important or very
improbable failures were skipped immediately and not put in
the table. Customers easily accept FMEA as a method for risk
assessment. The DRBFM approach has to be explained to
them thoroughly.
Very interesting is the emotional rating (how did you
feel?), which is strongly in favor of the DRBFM. Therefore, it
is a possibility to enhance the attention of the engineers. This
leads to better discussion and better results. Finally the overall
efficiency was also rated much more positive for the DRBFM
approach.
Summarizing the evaluation shows that the DRBFM
approach has advantages due to a faster possibility to reach
product improvements. This is for engineers more interesting

and worth spending time than to fill in an FMEA table. This


method is addressing more the creativity leading to more new
ideas. A disadvantage is the fact that possible failures are not
collected and summed up so completely as with an FMEA,
which therefore shows a lot of redundancy not necessary with
the DRBFM method. The FMEA is a widely known and
applied method showing high acceptance. If a quick but still
thorough risk assessment should be performed or a team
should improve a product or its reliability the DRBFM
approach appears to be the better choice.
This DRBFM approach can be summarized as follows:
1. Collect and prepare information on the prototype device
or change. This might include drawings, former products
or product information, concept images, material
properties, test results etc.
2. Put together the team. Next to product development
engineers, engineers from predevelopment or other
experts should be involved.
Additionally to the
moderator, it should consist of two or more product
engineers and two or more predevelopment engineers.
Their experience should cover the areas electrical,
thermal, mechanical, material issues or others.
3. Perform the DRBFM. Several shorter sessions are more
convenient than one long one.
Establish a nice
environment with coffee and snacks.
4. Summarize the results in a report and identify the
recommended actions. Present the results of the DRBFM,
discuss and define corrective actions. For future use the
discussion and results can also be documented in a
spreadsheet or database.
5 CONCLUSION
A DRBFM approach was used for risk assessment of a
power converter. It was evaluated and compared with
standard FMEA method. The risk evaluation using the
DRBFM method seems to be much more efficient, if the focus
lies on the identification of the most critical problems and on
product improvement. In contrast, FMEA is the method of
choice for assessments focusing at a complete list of possible
failures.
REFERENCES
1.

2.

3.
4.

P. K. Steimer, "Enabled by High Power Electronics Energy Efficiency, Renewables and Smart Grids",
International Power Electronics Conference, (Jun.) 2010,
pp. 11-15.
P. Barbosa, C. Haederli, P. Wilkstroem, M. Kauhanen, J.
Tolvanen, A. Savolainen, "Impact of Motor Drives on
Energy Efficiency", International Exhibition and
Conference for Power Electronics / Intelligent Motion /
Power Quality (PCIM), (May) 2007.
G. D. Demetriades, H. Z. de la Parra, "Trends in Power
Electronics and Variable Speed Drives", IEEE Power
Electronics Specialists Conference, (Jul.) 2007.
E. Echavarria, B. Hahn, G. J. W. van Bussel, T.
Tomiyama, "Reliability of Wind Turbine Technology
Through Time", Journal of Solar Energy Engineering,

5.
6.

7.
8.

9.

10.
11.
12.
13.
14.
15.
16.

17.

vol. 130, (Aug.) 2008.


J. Twidell, G. Gaudiosi (Eds.), Offshore Wind Power,
Multi-Science Publishing, 2009.
G. J. W. van Bussel, Offshore Wind Energy, the
Reliability Dilemma", Proceedings of the 1st World Wind
Energy Conference and Exhibition, Berlin, Germany,
(Jul.) 2002, pp 1-4.
W. Kanert, "Qualification Strategies in the Age of Zero
Defect". AEC Reliability Workshop 2006.
M. G. Pecht and A. Dasgupta, "Physics-of-Failure: An
Approach to Reliable Product Development", Journal of
the Institute of Environmental Sciences, vol. 38, 1995, pp.
30-34.
J. Snook, J. M. Marshall, R. M. Newman. "Physics of
Failure - As an Integrated Part of Design for Reliability".
Annual Reliability and Maintainability Symposium, (Jan.)
2003, pp. 46-54.
Tilo Pfeiffer, Quality Management: Strategies, Methods,
Techniques, Mnchen, Hanser Fachbuchverlag, 3rd
edition 2002.
D. H. Stamatis, Failure mode and effect analysis: FMEA
from theory to execution, Milwaukee, American Society
for Quality, 2nd edition, 2003.
N. Bidokhti, "FMEA is Not Enough", Proc. Ann.
Reliability & Maintainability Symp., (Jan.) 2009.
Z. Bluvband, P. Grabov, "Failure Analysis of FMEA",
Proc. Ann. Reliability & Maintainability Symp., (Jan.)
2009.
R. Schmidt, "Risk Assessment in Research Projects",
Proc. Ann. Reliability & Maintainability Symp., (Jan.)
2010.
B. Haughey, T. Yoshimura, "Design Review Based on
Failure Modes - DRBFM", International Applied
Reliability Symposium, San Diego, (Jun.) 2007.
H. Shimizu, T. Imagawa, H. Noguchi, "Reliability
Problem Prevention Method for Automotive Components
- Development of GD3 Activity and DRBFM (Design
Review Based on Failure Mode)", International Body
Engineering Conference, (Oct.) 2003, pp 371-376.
A. Kapust, Design Review Based on Failure Mode
(DRBFM), Training course of IMQ Consulting GmbH,
Koepffstrae 17, D-74076 Heilbronn, Germany,
www.tqm.com.
BIOGRAPHIES

Roland Schmidt
ABB Switzerland Ltd.
Corporate Research
Segelhofstrasse 1 K

Baden 5 Dttwil, 5405, Switzerland


e-mail: roland.schmidt@ch.abb.com
Roland Schmidt has more than 15 years experience with
reliability and quality. He works as principal scientist at
Corporate Research of ABB Switzerland on reliability of
power electronics. He is dealing with the reliability of
components, interconnection technologies and electrical
systems as well as quality and reliability methodologies. Prior
to joining ABB he worked on reliability and failure analytics
of semiconductor packages. He was also responsible for
supplier quality assurance in the semiconductors industry. He
received his diploma in physics from the University of
Erlangen and his PhD in physical chemistry for his work at the
Max-Planck-Institute for Polymer Research in Mainz.
Gernot J. Riedel
ABB Switzerland Ltd.
Corporate Research
Segelhofstrasse 1 K
Baden 5 Dttwil, 5405, Switzerland
e-mail: gernot.riedel@ch.abb.com
Gernot J. Riedel works as scientist at Corporate Research of
ABB Switzerland on topics regarding reliability in power
electronics since 2009. He mainly investigates the physics of
failures on component level and the reliability of systems. He
is also interested in system diagnostics and prognostics. He
received his PhD in physics for his work at the University of
Bristol, where he studied the thermal and electrical dynamics
of GaN based high frequency and high power electronic
devices. He received his diploma in physics from the
University of Karlsruhe (TH).
Klaus Kangas
ABB Oy Drives
Hiomotie 13
Helsinki, 00380, Finland
e-mail: klaus.kangas@fi.abb.com
Klaus Kangas has 8 years experience with reliability
engineering. He works as a reliability specialist at ABB Oy
Drives on topics regarding reliability engineering and
accelerated testing. Improving reliability through global ABB
reliability co-operation and projects is one part of his work.
Besides reliability topics in ABB, he has been for many years
a chairman of the reliability workgroup in Finnish national
KOTEL association. He received his diploma in electrical
engineering from the Helsinki University of Technology.

Das könnte Ihnen auch gefallen