DESIGN FAILURE MODE AND EFFECTS ANALYSIS (DFMEA)

The Design Failure Mode and Effects Analysis (Design FMEA) is a method for identifying potential or known failure modes and providing follow-up and corrective actions.

OBJECTIVE

The design FMEA is a disciplined analysis of the part design with the intent to identify and correct any known or potential failure modes before the manufacturing stage begins. Once these failure modes are identified and the cause and effects are determined, each failure mode is then systematically ranked so that the most severe failure modes receive priority attention. The completion of the design FMEA is the responsibility of the individual product design engineer. This individual engineer is the most knowledgeable about the product design and can best anticipate the failure modes and their corrective actions.

TIMING

The design FMEA is initiated during the early planning stages of the design and is continually updated as the program develops. The design FMEA must be totally completed prior to the first production run.

REQUIREMENTS

The requirements for a design FMEA include:

1. Forming a team

2. Completing the design FMEA form

3. FMEA risk ranking guidelines

DISCUSSION

The effectiveness of an FMEA is dependent on certain key steps in the analysis process, as follows :

Forming the Appropriate Team

A typical team for conducting a design FMEA is the following:

• Design engineer(s)

• Test/development engineer

• Reliability engineer

• Materials engineer

• Field service engineer

• Manufacturing/process engineer

• Customer

A design and a manufacturing engineer are required to be team members . Others may participate as needed or as the project calls for their knowledge or experience. The leader for the design FMEA is typically the design engineer.

Describing the Function of the Design/Product

There are three types of functions:

1. Task functions: These functions describe the single most important reason for the existence of the system/product. (Vacuum cleaner? Windshield wiper? Ballpoint pen?)

2. Supporting functions: These are the "sub" functions that are needed in order for the task function to be performed.

3. Enhancing functions: These are functions that enhance the product and improve customer satisfaction but are not needed to perform the task function.

After computing the function tree or a block diagram, transfer functions to the FMEA worksheet or some other form of a worksheet to retain. Add the extent of each function (range, target, specification, etc.) to test the measurability of the function.

Describing the Failure Mode Anticipated

The team must pose the question to itself, "How could this part, system or design fail? Could it break, deform, wear, corrode, bind, leak, short, open , etc.?" The team is trying to anticipate how the design being considered could possibly fail; at this point, it should not make the judgment as to whether it will fail but should concentrate on how it could fail.

The purpose of a design FMEA (DFMEA) is to analyze and evaluate a design on its ability to perform its functions. Therefore, the initial assumption is that parts are manufactured and assembled according to plan and in compliance with specifications.

• Once failure modes are determined under this assumption, then determine other failure modes due to purchased materials, components , manufacturing processes, and services.

Describing the Effect of the Failure

The team must describe the effect of the failure in terms of customer reaction or in other words, e.g., "What does the customer experience as a result of the failure mode of a shorted wire?" Notice the specificity. This is very important, because this will establish the basis for exploratory analysis of the root cause of the function. Would the shorted wire cause the fuel gage to be inoperative or would it cause the dome light to remain on?

Describing the Cause of the Failure

The team anticipates the cause of the failure. Would poor wire insulation cause the short? Would a sharp sheet metal edge cut through the insulation and cause the short? The team is analyzing what conditions can bring about the failure mode. The more specific the responses are, the better the outcome of the FMEA.

The purpose of a design FMEA (DFMEA) is to analyze and/or evaluate a design on its ability to perform its functions (part characteristics). Therefore, the initial assumption in determining causes is that parts are made and assembled according to plan and in compliance with specifications, including purchased materials, components, and services. Then and only then, determine causes due to purchased materials, components, and services.

Some cause examples include:

• Brittle material

• Weak fastener

• Corrosion

• Low hardness

• Too small of a gap

• Wrong bend angle

• Stress concentration

• Ribs too thin

• Wrong material selection

• Poor stitching design

• High G forces

• Part interference

• Tolerance stack-up

• Vibration

• Oxidation

• And so on

Estimating the Frequency of Occurrence of Failure

The team must estimate the probability that the given failure is going to occur. The team is assessing the likelihood of occurrence, based on its knowledge of the system, using an evaluation scale of 1 to 10. A 1 would indicate a low probability of occurrence whereas a 10 would indicate a near certainty of occurrence.

Estimating the Severity of the Failure

In estimating the severity of the failure, the team is weighing the consequence of the failure. The team uses the same 1 to 10 evaluation scale. A 1 would indicate a minor nuisance, while a 10 would indicate a severe consequence such as "loss of brakes" or "stuck at wide open throttle " or "loss of life."

Identifying System and Design Controls

Generally, these controls consist of tests and analyses that detect failure modes or causes during early planning and system design activities. Good system controls detect faults or weaknesses in system designs. Design controls consist of tests and analyses that detect failure causes or failure modes during design, verification, and validation activities. Good design controls detect faults or weaknesses in component designs.

Special notes:

• Just because there is a current control in place that does not mean that it is effective. Make sure the team reviews all the current controls, especially those that deal with inspection or alarms.

• To be effective (proactive), system controls must be applied throughout the pre-prototype phase of the Advanced Product Quality Planning (APQP) process.

• To be effective (proactive), design controls must be applied throughout the pre-launch phase of the APQP process.

• To be effective (proactive), process controls should be applied during the post-pilot build phase of APQP and continue during the production phase. If they are applied only after production begins, they serve as reactive plans and become very inefficient.

Examples of system and design controls include:

Engineering analysis

• Computer simulation

• Mathematical modeling/CAE/FEA

• Design reviews, verification, validation

• Historical data

• Tolerance stack studies

• Engineering reviews, etc.

System/component level physical testing

• Breadboard, analog tests

• Alpha and beta tests

• Prototype, fleet , accelerated tests

• Component testing (thermal, shock , life, etc.)

• Life/durability/lab testing

• Full scale system testing (thermal, shock, etc)

• Taguchi methods

• Design reviews

Estimating the Detection of the Failure

The team is estimating the probability that a potential failure will be detected before it reaches the customer. Again, the 1 to 10 evaluation scale is used. A 1 would indicate a very high probability that a failure would be detected before reaching the customer. A 10 would indicate a very low probability that the failure would be detected , and therefore, be experienced by the customer. For instance, an electrical connection left open preventing engine start might be assigned a detection number of 1. A loose connection causing intermittent no-start might be assigned a detection number of 6, and a connection that corrodes after time causing no start after a period of time might be assigned a detection number of 10.

Detection is a function of the current controls. The better the controls, the more effective the detection. It is very important to recognize that inspection is not a very effective control because it is a reactive task.

Calculating the Risk Priority Number

The product of the estimates of occurrence, severity, and detection forms a risk priority number (RPN). This RPN then provides a relative priority of the failure mode. The higher the number, the more serious is the mode of failure considered. From the risk priority numbers , a critical items summary can be developed to highlight the top priority areas where actions must be directed.

Recommending Corrective Action

The basic purpose of an FMEA is to highlight the potential failure modes so that the responsible engineer can address them after this identification phase. It is imperative that the team provide sound corrective actions or provide impetus for others to take sound corrective actions. The follow-up aspect is critical to the success of this analytical tool. Responsible parties and timing for completion should be designated in all corrective actions.

Strategies for Lowering Risk: (System/Design) ” High Severity or Occurrence

To reduce risk, you may change the product design to:

• Eliminate the failure mode cause or decouple the cause and effect

• Eliminate or reduce the severity of the effect

• Make the cause less likely or impossible to occur

• Eliminate function or eliminate part (functional analysis)

• Some "tools" to consider:

  • Quality Function Deployment (QFD)

  • Fault Tree Analysis (FTA)

  • Benchmarking

  • Brainstorming

  • TRIZ, etc.

• Evaluate ideas using Pugh concept selection. Some specific examples:

  • Change material, increase strength, decrease stress

  • Add redundancy

  • Constrain usage (exclude features)

  • Develop fail safe designs, early warning system

Strategies for Lowering Risk: (System/Design) ” High Detection Rating

Change the evaluation/verification/tests to:

• Make failure mode easier to perceive

• Detect causes prior to failure

• Some "tools" to consider:

  • Benchmarking

  • Brainstorming

  • Process control (automatic corrective devices)

  • TRIZ, etc.

• Evaluate ideas using Pugh concept selection. Some specific examples:

  • Change testing and evaluation procedures

  • Increase failure feedback or warning systems

  • Increase sampling in testing or instrumentation

  • Increase redundancy in testing