Failure Mode and Effects Analysis (FMEA): How to do it
The Quality Toolbook >
Failure Mode and Effects Analysis (FMEA) > How to do it
When to use it | How to understand it |
Example | How to do it | Practical
<-- Previous |
How to do it
- Select the item to be analyzed. If it is a part of another item, then be clear about the boundary. For example, if the item is 'vehicle doors', it may mean passenger doors, but not the tailgate.
- Identify the overall approach to be used. The FMEA may be a part of a larger set of failure analyses. In this case, the way that items are selected needs to be determined. Typical strategies include:
, where the system being analyzed is broken into pieces and FMEAs done on the larger items first. For example, starting with a whole vehicle and then successively breaking down into lower levels, such as doors, then catches, then screws.
, where the analyses of the smallest pieces are done first, followed by the higher level assemblies from which these are made. This is the reverse of top-down analysis.
, where the FMEAs are done on the physical parts of the system. This will typically use components specifications to determine failure levels.
, where the analysis is of the intended functions and operation of the system. This is looking at failure from the product user's standpoint, rather than the engineer's, and will typically use product specifications to determine failure modes.
Also decide whether to perform criticality analysis. This will require more effort, but will result in a numerical value being given to failure modes and effects, thus helping with prioritization of subsequent actions.
If doing criticality analysis, determine how it will be calculated. The method below focuses just on the probability of failure modes and effects, although this can be extended to account for other important factors, as indicated in the following section on Practical Variations.
Where possible, this method should utilize actual data, for example from product defect records. Otherwise define a range categories and corresponding numerical scores, then use an allocation method, such as Voting. For example, levels of 'Chance of being found in system test' being scored on a scale of 1 to 5.
- Identify the scope of failure to be examined. The scope defines the boundaries of the examination, and may include criteria such as time period, type of user, geography of use, etc.
For example, 'All vehicle doors failing to operate properly in final inspection and test for all shifts'.
- Design an appropriate table to capture the right information. This will vary, depending on factors such as if and how criticality is being measured, as
in the illustration.
- Identify items which may fail and which fall into the scope defined in step 2. This can be determined by asking, 'What can fail?' and may include individual components and any combinations, sub-assemblies, etc.
If this list becomes unmanageably large, then either reduce the scope of the FMEA, for example by examining just the catch mechanism rather than all parts of a door, or limit the detail of examination, for example by examining the catch mechanism as a whole, but not its individual components.
- If doing criticality analysis, determine the chance of failure for each item identified in step 5.
- If doing criticality analysis, identify the proportion of the time during the scope described in step 2 for each item identified in step 5 to fail. For example, if the scope is a defined test, then one item may only be exercised for 10% of the test time whilst another item is exercised for 90% of the time (and consequently has more opportunity to fail).
If all items may fail at any time, then this factor may be ignored, as it is always 100%.
- For each item identified in step 5, list all significant failure modes. These may be found by asking, 'How can it fail?'. For example, a hinge can seize, wear, fracture, etc. This can be simplified by identifying a standard list of failure modes for the item being examined.
- If doing criticality analysis, identify the chance of occurrence for each failure mode identified in step 8. If all possible failure modes for an item are identified, their chances of occurrence will total 100%.
- For each failure mode identified in step 8, determine all significant effects that may be manifested. Ask, 'What is an undesirable result of the identified failure mode?'. Again, this can be simplified by using a standard list of effects (e.g. won't close, difficult to close, stuck closed, etc.).
Note the difference between a failure mode and failure effect; a failure mode results in a failure effect. For example, a broken pedal may result in a cyclist falling off.
- If doing criticality analysis, identify the chance of occurrence of each failure effect identified in step 10. If all possible failure effects are identified, they will total 100% for each item.
- If doing criticality analysis, calculate the criticality of each failure effect identified in step 10, by multiplying together (a) the chance of the overall item failing (from step 6), the proportion of time that the item is at risk of failure (from step 7), the chance of the failure mode occurring (from step 9) and the chance of failure effect occurring (from step 10). This is illustrated
in the illustration.
- For each mode and effect that appears in more than one line in the table, usually because they are on a standard list, sum the criticality calculations from step 12 to determine its overall criticality rating.
Fig. 1. Building an FMEA table
- Examine the criticality scores and identify those failure modes and effects which will require action to be taken, and determine appropriate steps to reduce the chance of undesirable failure. Actions may include:
- Redesigning items to make them likely to fail.
- Adding items to handle failure of other items.
- Adding warning systems to alert when an item fails.
<-- Previous |