How to Get

More Out of

Your FMEAs


by Richard A. Harpster


Organizations have used the failure mode and effects analysis process for years. Recently, however, QS-9000's mandate that suppliers utilize FMEAs during new product design and development has shed new light on this useful tool. In addition to the automotive industry, many other industries, including aerospace and electronics, now use FMEAs.

Unfortunately, despite the countless training classes and numerous books written on its use, many organizations have failed to fully realize the benefits of the FMEA process. Informal surveys with more than 100 companies reveal that the majority of people using FMEAs do not perceive them as powerful tools but as something that must be done to meet a quality audit or customer requirement. One of the major reasons for this perception is that most FMEAs are performed and used incorrectly.

FMEAs can and should be a key element in any organization's advanced quality planning process. Organizations that properly use FMEAs have saved hundreds of thousands of dollars and report higher levels of customer satisfaction.

Understanding the differences between those who use and do not use FMEAs correctly can help your organization become one of those who experience the immense benefits that FMEAs can bring.

What are FMEAs?

Although several derivatives of FMEAs have been developed, there are two basic types: design and process.

Organizations utilize design FMEAs to verify that a product has been properly designed to meet all of the customer's requirements and that it can be manufactured at a target rate, cost and yield.

Design FMEAs capture the relationship between customer requirements, how a product can fail to meet the requirements, the effects of the failures, problems with the design that cause the failures and how the design will be validated to prove it will not fail. Rating columns exist for the effects of failure, probability of failure and the effectiveness of the design validation. These ratings are multiplied together to achieve a risk priority number (RPN) for each product failure and cause combination. Rating numbers typically range from one to five or one to 10 for each of the columns, with higher numbers designating unacceptable conditions.

A class column serves to identify design characteristics that require special attention. An improvement section exists to identify and track attempts to improve the design.

Organizations utilize process FMEAs to assess the adequacy of a process in producing the product whose design the design FMEA has validated. Process FMEAs also identify the process and product controls that must be implemented to ensure that the product can be produced within specifications.

Process FMEAs capture the relationship between each process step, the unacceptable process outputs that can be created at each step, the effects of the unacceptable process outputs, causes of the unacceptable outputs and how the unacceptable outputs will be either prevented or detected in the event they occur. Rating columns exist for the effects or severity of the unacceptable process outputs, probability of outputs occurring and the effectiveness of the prevention and detection methods. As with the design FMEA, these ratings are multiplied together to achieve an RPN for each unacceptable output and cause combination.

A class column serves to identify design characteristics that require special attention. An improvement section exists to identify and track attempts to improve the process.

For both types of FMEAs, RPNs above a certain level trigger recommended actions to address the problem.

Successful and unsuccessful FMEA implementation

Organizations that use FMEAs can normally be categorized in one of five implementation stages. Organizations that achieve measurable monetary benefits operate in stages four or five. Unfortunately, most organizations remain in stages one through three, with most in stage one.

Stage one

Typically, the organization in stage one uses FMEAs because it has to meet a paper requirement for a customer or quality standard. Personnel perform the FMEA right before it is due to be turned in to the customer, usually too late in the process to be useful.

Quite often, the wrong people perform the FMEA. The quality department ends up developing the documents rather than making design engineers responsible for design FMEAs and operating personnel responsible for process FMEAs.

Management does not understand the FMEA, and a lot of confusion and disagreement exists as to how to fill out the FMEA form. Debate occurs when individuals attempt to develop the ratings for occurrence and detection numbers. Because accurate feedback systems don't yet exist to base the ratings on, they're based on inaccurate guesses. Using inaccurate ratings, the organization calculates erroneous RPNs and identifies an RPN level at which recommended actions are needed. If the number of recommended actions required based on this level is too high, the organization "adjusts" the ratings to bring the RPNs down below the trigger level, which reduces the number of recommended actions. Obviously, this makes the entire FMEA process meaningless.

As a result, the organization fulfills its FMEA paper requirement, but the value of the FMEA is greatly diminished. Problems still remain unsolved at a high cost to the organization. The individuals performing the FMEAs believe they are doing them correctly because the customer or auditor is accepting them. Eventually, everyone sees the FMEA process not as a tool but as something that has to be done.

Stage two

Management ensures that the personnel who will perform and use the FMEA data are trained in the proper technique. They realize that the people who perform design FMEAs must be experts in the product and that the people who perform process FMEAs must be experts in the process. Rather than being confused by the FMEA terminology, they realize they have used the FMEA methodology before but never called it FMEA. They also learn that although they have used the methodology, they have not used it rigorously enough to achieve its full benefits.

In stage two, everyone involved gains an understanding of what the ratings and class column mean and how to use them to prioritize what must be worked on first. They understand that the class column is the most important factor and not the RPN. (See "The Problem with Risk Priority Numbers.") Management also realizes that they don't have systems in place that will give them data to accurately determine the failure probability occurrence ratings, detection ratings and class. Using the limited objective data they have, they know they will have to use their knowledge of the product and process to arrive at the ratings. Due to the lack of an objective basis, they know that it is a waste of time to argue for long periods about the ratings.

When leaving this stage, those who have been using FMEAs believe they can be a powerful tool. Unfortunately, the people who perform the FMEAs doubt whether management will provide the time and resources necessary to support their successful implementation. They also question how they will explain the new approach to their auditor or customer, who may still be at stage one in their understanding of the FMEA implementation process.

Stage three

The organization begins to use FMEAs correctly on a targeted product. Early on there is excitement that the FMEAs are finally going to be done correctly. As the implementation continues, worry starts to set in as the FMEA uncovers and documents the complexity of the product and process being analyzed. Everyone knew the complexity existed but had never seen it documented. The FMEA grows from the five to 10 pages that used to be normal to 100 or more pages. The organization must overcome its fear of the increased length and complexity of the complete FMEA if FMEAs are to be used successfully. "FMEAs can and should be a key element in any organization's advanced quality planning process."

As the FMEA process continues, many problems may be uncovered that must be solved if the company is to become as good as it can be. There may not be enough resources to solve all of these problems and still meet the launch deadline.

Knowing this, people begin to proclaim that FMEAs will never work. They believe that all the hard work has been a waste of time. What good is it to know what is wrong and not be able to correct it? Life was a lot easier when all the problems were not documented. If organizations don't overcome this obstacle they may slip back into stage one.

Stage four

Management realizes that the length of an FMEA cannot be predetermined. The complexity of the product and process being analyzed determines the FMEA's length. Management understands that all of the problems uncovered in the FMEA can't be solved in one product launch. They understand that the product will be launched with known problems and that they will have to make objective decisions as to what to work on during this launch and what must be delayed. When a problem occurs in an area that management decided not to work on, it is handled without emotion. This is possible because the FMEA identified that it might happen and management chose not to work on it so that other problems could be prevented.

Once the organization launches the product, management creates a long-term plan to improve the design and manufacturing systems. This plan minimizes the times the company must make difficult decisions when doing FMEAs in the future.

Stage five

The organization has implemented new design and manufacturing systems to answer the majority of the problems identified in the FMEAs. Systems now exist to provide data to accurately set occurrence and detection ratings. Due to the accuracy of the new ratings, predictions about field failures and process yields can be made. The class column can now be accurately determined and actions required for improvement prioritized.

Design engineers review design FMEAs before making design changes. If a change must be made, operating personnel review the process FMEA and control plan to determine the impact the change will have on the process.

When a problem occurs, the appropriate personnel consult the FMEAs. If the FMEA inadequately addressed the problem, engineers make changes to the design and manufacturing system to ensure that all possible steps have been taken to prevent a similar problem in the future.

The organization uses the FMEAs as training tools because they contain the collective knowledge of the company's experts.


The FMEA process can be a powerful tool when properly used. As with any tool, before it can be used well it must be understood. Once organizations fully understand and commit to the FMEA process, they may be pleasantly surprised with the money they save and the product and process improvements that result.

About the author

Richard A. Harpster has more than 24 years' experience in engineering, operations, maintenance and quality. After 14 years at Ford, he founded Harpco Systems Inc. a training, consulting and software firm that specializes in the implementation of FMEAs. The company's Quality Plus software product automatically writes FMEAs and is used by both Fortune 500 and smaller companies. Harpster can be contacted at (248) 669-3706 or at .

1999 Harpco Systems Inc.

The Problem with Risk Priority Numbers

Almost everyone who uses FMEAs is taught to use the RPN to determine on which line of the FMEA to work. Customers tell their suppliers they want recommended actions written for all lines with RPNs over a certain value. Auditors ask auditees for the RPN level that triggers the need for a recommended action. The following example demonstrates the problems with this approach.

Figure 1:
Two sets of ratings from two lines of a process FMEA.

Severity    Occurrence     Detection     Risk Priority #

   8                    5                    1                     40
   8                    1                  10                     80

Using the automotive industry's generally accepted rating definitions, the first line in Figure 1 indicates the following: The severity rating of 8 means that the unacceptable process output will result in a loss of product function. The occurrence rating of 5 indicates that the process failure cause for the line will produce the unacceptable output at a probability between 1/400 and 1/2,000. The detection rating of 1 indicates that inspection is available in the process to successfully sort out the defect.

The second line indicates the following: The severity rating of 8 means that the unacceptable process output will result in a loss of product function. The occurrence rating of 1 indicates that the process failure cause for the line will produce the unacceptable output at a probability less than 1/1,500,000. Since the probability of causing the defect is so rare, the operating personnel have chosen not to try to prevent the cause of the defect or inspect for it. Since there are no process controls available, the detection rating must be a 10.

A quick review of the RPNs shows that if an RPN of 50 (typical) was used as a guideline, recommended actions would be developed to either prevent a cause that rarely results in a defect or inspect for a defect that rarely happens. All this will be done while the process is rejecting between 1/400 to 1/2,000 of its products without any signs that action is needed. When users of FMEAs are guided by the document to work on items that are logically incorrect, they lose respect for the FMEA process.

Those who use FMEAs need to learn that the class value is more important than the RPN. The class value, derived from the severity/occurrence matrix shown in Figure 2 below, puts the RPN into its proper perspective. The class matrix shows that any FMEA line with a severity rating of nine or 10 lies within the Legal Zone and poses a possible legal risk. These must be addressed regardless of the occurrence rating. The Warranty Zone shows combinations of severity and occurrence ratings that could lead to field failure with a probability that is unacceptable to the company (typically > 1/15,000). These items also need to be addressed.

In the example in Figure 1, the RPN of 40 seems to indicate no need for action. However, the class matrix indicates that line one falls into the Warranty Zone, indicating a problem that will lead to dissatisfied customers and cost the company money.

Line two in Figure 1 shows an RPN of 80 that would raise flags for an auditor but doesn't fall into either the Legal Zone or Warranty Zone. Therefore, careful consideration must be given as to whether to spend time and money correcting this problem.

Unfortunately, auditors and those who use FMEAs often look only at the RPN and ignore the class, which leads to wasted effort on problems of little significance.

Figure 2: Class matrixFmea

[QD Online] [Buyers Guide] [Information] [ISO 9000] [CMM] [FMEA] [Video]

Copyright 1999 QCI International. All rights reserved. Quality Digest can be reached by phone at (530) 893-4095. E-mail: Click Here