There are many forms to develop a typical FMEA. However, all of them are basically the same in that they are made up of two parts , whether they are for system, design, process, or machinery. A typical FMEA form consists of the header information and the main body.
There is no standard information that belongs in the header of the form, but there are specific requirements for the body of the form.
In the header, one may find the following information ” see Figure 6.7. However, one must remember that this information may be customized to reflect one's industry or even the organization:
Type of FMEA study
Subject description
Responsible engineer
FMEA team leader
FMEA core team members
Suppliers
Appropriate dates (original issue, revision, production start, etc.)
FMEA number
Assembly/part/detail number
Current dates (drawings, specifications, control plan, etc.)
The form may be expanded to include or to be used for such matters as:
Safety: Injury is the most serious of all failure effects. As a consequence, safety is handled either with an FMEA or a fault tree analysis (FTA) or critical analysis (FMCA). In the traditional FTA, the starting point is the list of hazard or undersized events for which the designer must provide some solution. Each hazard becomes a failure mode and thus it requires an analysis.
Figure 6.6: Scope for PFMEA ” printed circuit board screen printing process.
Figure 6.7: Typical FMEA header.
Effect of downtime: The FMEA may incorporate maintenance data to study the effects of downtime. It is an excellent tool to be used in conjunction with total preventive maintenance.
Repair planning: An FMEA may provide preventive data to support repair planning as well as predictive maintenance cycles.
Access: In the world of recycling and environmental conscience, the FMEA can provide data for tear downs as well as information about how to get at the failed component. It can be used with mistake proofing for some very unexpected positive results.
A typical body of an FMEA form may look like Figure 6.8. The details of this form will be discussed in the following pages. We begin with the first part of the form; that is the description in the form of:
Part name /process step and function (verb/noun) |
In this area the actual description is written in concise , exact and simple language. |
A fundamental principle in writing functions is the notion that they must be written either in action verb format or as a measurable noun. Remember, a function is a task that a component, subsystem, or product must perform, described in language that everyone understands. Stay away from jargon. To identify appropriate functions, leading questions such as the following may help:
What does the product/process do?
How does the product/process do that?
If a product feature or process step is deleted, what functions disappear?
If you were this task, what are you supposed to accomplish? Why do you exist?
The priority of asking function questions for a system/part FMEA is:
A system view
A subsystem view
A component view
Typical functions are:
Position
Support
Seal in, out
Retain
Lubricate
After the brainstorming is complete, a function tree ” see Figure 6.9 ” can be used to organize the functions. This is a simple tree structure to document and help organize the functions, as follows :
Purposes of the function tree
To document all the functions
To improve team communication
To document complexity and improve team understanding of all the functions
Steps
Brainstorm all the functions.
Arrange functions into function tree.
Test for completeness of function (how/why).
Building the function tree
Ask:
What does the product/process do?
Which component/process step does that?
How does it do that?
Primary functions provide a direct answer to this question without conditions or ambiguity.
Secondary functions explain how primary functions are performed.
Continue until the answer to "how" requires using a part name, labor operation, or activity.
Ask "why" in the reverse direction.
Add additional functions as needed.
The function tree process can be summarized as follows:
Identify the task function.
Place on the far left side of a chart pad.
Identify the supporting functions.
Place on the top half of the pad.
Identify enhancing functions.
Place on the bottom half of the pad.
Build the function tree.
Include the secondary/ tertiary functions.
Place these to the right of the primary functions.
Verify the diagram: Ask how and why.
For an example of a function tree for a ball point pen (tip), see Figure 6.10.
The second portion of the FMEA body form deals with the failure mode analysis. A typical format is shown in Figure 6.11.
Failure mode (a specific loss of a function) is the inability of a component/subsystem/system/process/part to perform to design intent. In other words, it may potentially fail to perform its function(s). Design failure mode is a technical description of how the system, subsystem, or part may not adequately perform its function. Process failure mode is a technical description of how the manufacturing process may not perform its function, or the reason the part may be rejected.
The process of brainstorming failure modes may include the following questions:
DFMEA
Considering the conditions in which the product will be used, how can it fail to perform its function?
How have similar products failed in the past?
PFMEA
Considering the conditions in which the process will be used, what could possibly go wrong with the process?
How have similar processes failed in the past?
What might happen that would cause a part to be rejected?
Failure modes are when the function is not fulfilled in five major categories. Some of these categories may not apply. As a consequence, use these as "thought provokers" to begin the process and then adjust them as needed:
Absence of function
Incomplete, partial, or decayed function
Related unwanted "surprise" failure mode
Function occurs too soon or too late
Excess or too much function
Interfacing with other components , subsystems or systems. There are four possibilities of interfacing. They are (a) energy transfer, (b) information transfer, (c) proximity, and (d) material compatibility.
Failure mode examples using the above categories and applied to the pen case include:
Absence of function:
DFMEA: Make marks
PFMEA: Inject plastic
Incomplete, partial or decayed function:
DFMEA: Make marks
PFMEA: Inject plastic
Related unwanted "surprise" failure mode
DFMEA: Make marks
PFMEA: Inject plastic
Function occurs too soon or too late
DFMEA: Make marks
PFMEA: Inject plastic
Excess or too much function
DFMEA: Make marks
PFMEA: Inject plastic
General examples of failure modes include:
Design FMEA:
No power | Failed to open |
Water leaking | Partial insulation |
Open circuit | Loss of air |
Releases too early | No spark |
Noise | Insufficient torque |
Vibration | Paper jams |
Does not cut | And so on |
Process FMEA:
Four categories of process failures:
Fabrication failures
Assembly failures
Testing failures
Inspection failures
Typical examples of these categories are:
Warped
Too hot
RPM too slow
Rough surface
Loose part
Misaligned
Poor inspection
Hole too large
Leakage
Fracture
Fatigue
And so on
Note | At this stage, you are ready to transfer the failure modes in the FMEA form ” see Figure 6.12. Figure 6.12: Transferring the failure modes to the FMEA form. |
A failure mode effect is a description of the consequence/ramification of a system, part, or manufacturing process failure. A typical failure mode may have several effects depending on which customer(s) are considered . Consider the effects/consequences on all the "customers," as they are applicable , as in the following FMEAs:
SFMEA
System
Other systems
Whole product
Government regulations
End user
DFMEA
Part
Higher assembly
Whole product
Government regulations
End user
PFMEA
Part
Next operation
Equipment
Government regulations
Operators
End user
Effects and severity are very related items. As the effect increases , so does the severity. In essence, two fundamental questions have to be raised and answered :
What will happen if this failure mode occurs?
How will customers react if these failures happen?
Describe as specifically as possible what the customer(s) might notice once the failure occurs.
What are the effects of the failure mode?
How severe is the effect on the customers?
The progression of function, cause, failure mode, effect, and severity can be illustrated by the following series of questions:
In function: What is the individual task intended by design?
In failure mode: What can go wrong with this function?
In cause: What is the "root cause" of the failure mode?
In effect: What are the consequences of this failure mode?
In severity: What is the seriousness of the effect?
The following are some examples of DFMEA and PFMEA effects:
Customer gets wet | Loss of performance |
System failure | Scrap |
Loss of efficiency | Rework |
Reduced life | Becomes loose |
Degraded performance | Hard to load in next operation |
Cannot assemble | Operator injury |
Violate Gov. Reg. XYZ | Noise, rattles |
Damaged equipment | And so on |
Special Note: | Please note that the effect remains the same for both DFMEA and PFMEA. |
Severity is a numerical rating ” see Table 6.1 for design and Table 6.2 for process ” of the impact on customers. When multiple effects exist for a given failure mode, enter the worst-case severity on the worksheet to calculate risk. (This is the excepted method for the automotive industry and for the SAE J1739 standard. In cases where severity varies depending on timing, use the worst-case scenario.
Effect | Description | Rating |
---|---|---|
None | No effect noticed by customer; the failure will not have any perceptible effect on the customer | 1 |
Very minor | Very minor effect, noticed by discriminating customers; the failure will have little perceptible effect on discriminating customers | 2 |
Minor | Minor effect, noticed by average customers; the failure will have a minor perceptible effect on average customers | 3 |
Very low | Very low effect, noticed by most customers; the failure will have some small perceptible effect on most customers | 4 |
Low | Primary product function operational, however at a reduced level of performance; customer is somewhat dissatisfied | 5 |
Moderate | Primary product function operational, however secondary functions inoperable; customer is moderately dissatisfied | 6 |
High | Failure mode greatly affects product operation; product or portion of product is inoperable; customer is very dissatisfied | 7 |
Very high | Primary product function is non-operational but safe; customer is very dissatisfied. | 8 |
Hazard with warning | Failure mode affects safe product operation and/or involves nonconformance with government regulation with warning | 9 |
Hazard with no warning | Failure mode affects safe product operation and/or involves nonconformance with government regulation without warning | 10 |
Effect | Description | Rating |
---|---|---|
None | No effect noticed by customer; the failure will not have any effect on the customer | 1 |
Very minor | Very minor disruption to production line; a very small portion of the product may have to be reworked; defect noticed by discriminating customers | 2 |
Minor | Minor disruption to production line; a small portion (much <5%) of product may have to be reworked on-line; process up but minor annoyances | 3 |
Very low | Very low disruption to production line; a moderate portion (<10%) of product may have to be reworked on-line; process up but minor annoyances | 4 |
Low | Low disruption to production line; a moderate portion (<15%) of product may have to be reworked on-line; process up but minor annoyances | 5 |
Moderate | Moderate disruption to production line; a moderate portion (>20%) of product may have to be scrapped; process up but some inconveniences | 6 |
High | Major disruption to production line; a portion (>30%) of product may have to be scrapped; process may be stopped ; customer dissatisfied | 7 |
Very high | Major disruption to production line; close to 100% of product may have to be scrapped; process unreliable; customer very dissatisfied | 8 |
Hazard with warning | May endanger operator or equipment; severely affects safe process operation and/or involves noncompliance with government regulation; failure will occur with warning | 9 |
Hazard with no warning | May endanger operator or equipment; severely affects safe process operation and/or involves noncompliance with government regulation; failure occurs without warning | 10 |
Note | There is nothing special about these guidelines. They may be changed to reflect the industry, the organization, the product/design, or the process. For example, the automotive industry has its own version and one may want to review its guidelines in the AIAG (2001). To modify these guidelines, keep in mind:
|
At this point the information should be transferred to the FMEA form ” see Figure 6.13. The column identifying the "class" is the location for the placement of the special characteristic. The appropriate response is only "Yes" or "No." A Yes in this column indicates that the characteristic is special, a No indicates that the characteristic is not special. In some industries, special characteristics are of two types: (a) critical and (b) significant. "Critical" refers to characteristics associated with safety and/or government regulations, and "significant" refers to those that affect the integrity of the product. In design, all special characteristics are potential. In process they become critical or significant depending on the numerical values of severity and occurrence combinations.
The analysis of the cause and occurrence is based on two questions:
What design or process choices did we already make that may be responsible for the occurrence of a failure?
How likely is the failure mode to occur because of this?
For each failure mode, the possible mechanism(s) and cause(s) of failure are listed. This is an important element of the FMEA since it points the way toward preventive/corrective action. It is, after all, a description of the design or process deficiency that results in the failure mode. That is why it is important to focus on the "global" or "root" cause. Root causes should be specific and in the form of a characteristic that may be controlled or corrected. Caution should be exerted not to overuse "operator error" or "equipment failure" as a root cause even though they are both tempting and make it easy to assign "blame."
You must look for causes, not symptoms of the failure. Most failure modes have more than one potential cause. An easy way to probe into the causes is to ask:
What design choices, process variables , or circumstances could result in the failure mode(s)?
DFMEA failure causes are typically specific system, design, or material characteristics.
PFMEA failure causes are typically process parameters, equipment characteristics, or environmental or incoming material characteristics.
Ways to determine failure causes include the following:
Brainstorm
Whys
Fishbone diagram
Fault Tree Analysis (FTA; a model that uses a tree to show the cause-and-effect relationship between a failure mode and the various contributing causes. The tree illustrates the logical hierarchy branches from the failure at the top to the root causes at the bottom.)
Classic five-step problem-solving process
What is the problem?
What can I do about it?
Put a star on the "best" plan.
Do the plan.
Did your plan work?
Kepner Trego (What is, what is not analysis)
Discipline GPS - see Volume II
Experience
Knowledge of physics and other sciences
Knowledge of similar products
Experiments ” When many causes are suspect or specific cause is unknown
Classical
Taguchi methods
The occurrence rating is an estimated number of frequencies or cumulative number of failures (based on experience) that will occur in our design concepts for a given cause over the intended life of the design. For example: cause of staples falling out = soft wood. The likelihood of occurrence is a 9 if we pick balsa wood but a 2 if we choose oak.
Just as with severity, there are standard tables for occurrence ” see Table 6.3 for design and Table 6.4 for process ” for each type of FMEA. The ratings on these tables are estimates based on experience or similar products or processes. Non-standard occurrence tables may also be used, based on specific characteristics. However, reliability expertise is needed to construct occurrence tables. (Typical characteristics may be historical failure frequencies, C pks , theoretical distributions, and reliability statistics.)
Occurrence | Description | Frequency | Rating |
---|---|---|---|
Remote | Failure is very unlikely | < 1 in 1,500,000 | 1 |
Low | Relatively few failures | 1 in 150,000 | 2 |
1 in 15,000 | 3 | ||
Moderate | Occasional failures | 1 in 2000 | 4 |
1 in 400 | 5 | ||
1 in 80 | 6 | ||
High | Repeated failures | 1 in 20 | 7 |
1 in 8 | 8 | ||
Very high | Failure is almost inevitable | 1 in 3 | 9 |
>1 in 2 | 10 |
Occurrence | Description | Frequency | C pk | Rating |
---|---|---|---|---|
Remote | Failure is very unlikely; no failures associated with similar processes | <1 in 1,500,000 | >1.67 | 1 |
Low | Few failures; isolated failures associated with like processes | 1 in 150,000 1 in 15,000 | 1.50 1.33 | 2 3 |
Moderate | Occasional failures associated with similar processes, but not in major proportions | 1 in 2,000 1 in 400 1 in 80 | 1.17 1.00 0.83 | 4 5 6 |
High | Repeated failures; similar processes have often failed | 1 in 20 1 in 8 | 0.67 | 7 8 |
Very high | Process failure is almost inevitable | 1 in 3 >1 in 2 | 0.51 0.33 | 9 10 |
At this point the data for causes and their ratings should be transferred to the FMEA form ” see Figure 6.14.
Design and process controls are the mechanisms, methods, tests, procedures, or controls that we have in place to prevent the cause of the failure mode or detect the failure mode or cause should it occur. (The controls currently exist.)
Design controls prevent or detect the failure mode prior to engineering release.
Process controls prevent or detect the failure mode prior to the part or assembly leaving the area.
A good control prevents or detects causes or failure modes.
As early as possible ( ideally before production or prototypes )
As early as possible
Using proven methods
So, the next step in the FMEA process is to:
Analyze planned controls for your system, part, or manufacturing process
Understand the effectiveness of these controls to detect causes or failure modes
Detection rating ” see Table 6.5 for design and Table 6.6 for process ” is a numerical rating of the probability that a given set of controls will discover a specific cause or failure mode to prevent bad parts from leaving the operation/facility or getting to the ultimate customer. Assuming that the cause of the failure did occur, assess the capabilities of the controls to find the design flaw or prevent the bad part from leaving the operation/facility. In the first case, the DFMEA is at issue. In the second case, the PFMEA is of concern.
Detection | Description | Rating |
---|---|---|
Almost certain | Design control will almost certainly detect the potential cause of subsequent failure modes | 1 |
Very high | Very high chance the design control will detect the potential cause of subsequent failure mode | 2 |
High | High chance the design control will detect the potential cause of subsequent failure mode | 3 |
Moderately high | Moderately high chance the design control will detect the potential cause of subsequent failure mode | 4 |
Moderate | Moderate chance the design control will detect the potential cause of subsequent failure mode | 5 |
Low | Low chance the design control will detect the potential cause of subsequent failure mode | 6 |
Very low | Very low chance the design control will detect the potential cause of subsequent failure mode | 7 |
Remote | Remote chance the design control will detect the potential cause of subsequent failure mode | 8 |
Very remote | Very remote chance the design control will detect the potential cause of subsequent failure mode | 9 |
Very uncertain | There is no design control or control will not or cannot detect the potential cause of subsequent failure mode | 10 |
Detection | Description | Rating |
---|---|---|
Almost certain | Process control will almost certainly detect or prevent the potential cause of subsequent failure mode | 1 |
Very high | Very high chance process control will detect or prevent the cause of subsequent failure mode | 2 |
High | High chance the process control will detect or prevent the potential cause of subsequent failure mode. | 3 |
Moderately high | Moderately high chance the process control will detect or prevent the potential cause of subsequent failure mode | 4 |
Moderate | Moderate chance the process control will detect or prevent the potential cause of subsequent failure mode | 5 |
Low | Low chance the process control will detect or prevent the potential cause of subsequent failure mode | 6 |
Very low | Very low chance the process control will detect or prevent the potential cause of subsequent failure mode | 7 |
Remote | Remote chance the process control will detect or prevent the potential cause of subsequent failure mode | 8 |
Very remote | Very remote chance the process control will detect or prevent the potential cause of subsequent failure mode | 9 |
Very uncertain | There is no process control or control will not or cannot detect the potential cause of subsequent failure mode | 10 |
When multiple controls exist for a given failure mode, record the best ( lowest ) to calculate risk. In order to evaluate detection, there are appropriate tables for both design and process. Just as before, however, if there is a need to alter them, remember that the change and approval must be made by the FMEA team with consensus.
At this point, the data for current controls and their ratings should be transferred to the FMEA form ” see Figure 6.15. There should be a current control for every cause. If there is not, that is a good indication that a problem might exist.
Without risk, there is very little progress. Risk is inevitable in any system, design, or manufacturing process. The FMEA process aids in identifying significant risks, then helps to minimize the potential impact of risk. It does that through the risk priority number or as it is commonly known, the RPN index. In the analysis of the RPN, make sure to look at risk patterns rather than just a high RPN. The RPN is the product of severity, occurrence, and detection or:
Risk = RPN = S — O — D
Obviously the higher the RPN the more the concern. A good rule-of-thumb analysis to follow is the 95% rule. That means that you will address all failure modes with a 95% confidence. It turns out the magic number is 50, as indicated in this equation: [(S = 10 — O = 10 — D = 10) - (1000 — .95)]. This number of course is only relative to what the total FMEA is all about, and it may change as the risk increases in all categories and in all causes.
Special risk priority patterns require special attention, through specific action plans that will reduce or eliminate the high risk factor. They are identified through:
High RPN
Any RPN with a severity of 9 or 10 and occurrence > 2
Area chart
The area chart ” Figure 6.16 ” uses only severity and occurrence and therefore is a more conservative approach than the priority risk pattern mentioned previously. At this stage, let us look at our FMEA project and calculate and enter the RPN ” see Figure 6.17. It must be noted here that this is only one approach to evaluating risk. Another possibility is to evaluate the risk based on the degree of severity first, in which case the engineer tries to eliminate the failure; evaluate the risk based on a combination of severity (values of 5 “8) and occurrence (>3) second, in which case the engineer tries to minimize the occurrence of the failure through a redundant system; and to evaluate the risk through the detection of the RPN third, in which case the engineer tries to control the failure before the customer receives it.